This blog post was originally published at Ambarella’s website. It is reprinted here with the permission of Ambarella.

Thanks to the commercialization of robotaxis, the driverless revolution is almost here.

Modern autonomy is understood as a range of capabilities laid out on a scale—level 0 through 5. On the lower end (levels 1 and 2), autonomous simply means that one or more driving behaviors, such as steering or braking, are capable of being performed automatically under strictly defined conditions—and only while the driver remains attentive, hands on the wheel, ready to override control. At level 3, we see the first instance of true hands-off driving, as the vehicle becomes capable of performing all maneuvers within the boundaries of geofenced metropolitan areas in limited scenarios, with a driver ready to assume command at the vehicle’s request. At level 4, the geofencing requirement remains, but there is no longer any expectation of human intervention, while at level 5 (the highest level of autonomy), all boundaries and restrictions disappear. Not only is a driver unnecessary, but the vehicle is no longer confined to carefully managed environments; rather, it becomes truly independent: aware, informed, and perceiving.

The implementation of autonomy at levels 4 and 5—making the dream of driverless mobility possible—is a tremendous undertaking. It starts with extensive testing on closed courses and public roadways in hundreds of challenging and unpredictable scenarios involving other vehicles, pedestrians, cyclists, intersections, roundabouts, varying weather conditions, difficult lighting, and more. But the challenge doesn’t end there. Even if these obstacles can be met from a technological perspective, even if every test can be passed on paper, the next question becomes: does the driverless technology have a clear path to market? Is it deployable at scale?

Implementing driverless mobility using experimental prototypes is only the beginning. Because these vehicles are typically inefficient and impractical by design, outfitted with expensive sensor suites characterized by high computational resource requirements, the road to commercialization isn’t always straightforward. It requires a number of very challenging prerequisites: (1.) maintaining strong AI processing performance, (2.) lowering cost for volume production, (3.) satisfying the power consumption demands of real-world use, and (4.) achieving functional safety certification, a rigorous process designed to ensure that automotive technologies are ready for the road.

Our colleagues at VisLab (acquired by Ambarella in 2015) understood these realities as early as the 1990s. Globally recognized as pioneers in vehicular autonomy, VisLab spent decades developing a series of groundbreaking self-driving prototypes that helped shape the evolution of the industry. And as they achieved these level 4 and level 5 milestones, they accomplished something else as well—they pushed autonomy toward practicality. Performance was their first priority, but it never came at the expense of usability and scalability. They always envisioned their creations freed from the lab, deployed in the real world, and they implemented their solutions with this overarching vision in mind.

The robotaxi revolution

As autonomous research progressed, robotaxis became one of the prime catalysts for the commercialization of driverless technologies. These vehicles offer a perfect application for level 4 autonomy: taxis operate within a defined geographic area, and the lower cost per mile of autonomous vehicles is especially attractive for the ridesharing industry. As a result, manufacturers have developed several scalable platforms for deploying fully autonomous vehicles in volumes high enough to fill commercial taxi fleets worldwide.

Let’s look at Motional, for instance. Motional is a joint venture between Hyundai Motor Group, one of the world’s largest vehicle manufacturers offering smart mobility solutions, and Aptiv, one of the industry’s most innovative technology providers. Motional is a driverless technology pioneer, responsible for many industry firsts, including being one of the first companies in the world to operate driverless vehicles on public roads. In addition, the company launched the world’s first robotaxi pilot, and currently operates the longest-standing commercial robotaxi service. Motional has provided more than 100,000 rides to the public, with zero at-fault incidents to date. And in December 2020, the company announced a landmark agreement with Lyft for fully driverless deployments in multiple cities across the Lyft network.

We at Ambarella are proud that our CVflow SoCs were selected to be part of the central processing module in each of Motional’s driverless vehicles, providing image and computer vision processing for cameras in the sensing suite, including the front-facing cameras. Our CVflow AI engine will enable Motional AI algorithms to perform complex computer vision tasks, such as object detection, classification, and image segmentation, with industry-leading power efficiency. In addition, our advanced image processing will allow the vehicles to operate in challenging lighting conditions, including low-light and high-contrast scenarios, while our chips’ H.264 encoding capabilities will enable the efficient logging of video data from all cameras in the vehicle.

For more information on our level 4 and 5 autonomous demonstrations, please contact us.

Senya Pertsel
Senior Director, Marketing, Ambarella

The post The Rise of Level 4 and Level 5 Autonomy appeared first on Edge AI and Vision Alliance.

LETTER FROM THE EDITOR

Dear Colleague, Until this Friday, January 28, the Edge AI and Vision Alliance is accepting applications for the 2022 Edge AI and Vision Product of the Year Awards competition. The Edge AI and Vision Product of the Year Awards celebrate the innovation of the industry’s leading companies that are developing and enabling the next generation of edge AI and computer vision products. Winning a Product of the Year award recognizes your leadership in edge AI and computer vision as evaluated by independent industry experts. Winners will be publicly announced at the Embedded Vision Summit, the key event for system and application developers who are incorporating computer vision and visual AI into products. For more information on the Edge AI and Vision Product of the Year Awards and to enter, please see the program page. Again, the deadline for applications is January 28. Registration for the Embedded Vision Summit, taking place May 17-19 in Santa Clara, California, is also now open, and if you register by March 11, you can save 25% by using the code SUMMIT22-NL. On Thursday February 24 at 9 am PT, BrainChip will deliver the free webinar “Developing Optimized Systems with BrainChip’s Akida Neuromorphic Processor” in partnership with the Edge AI and Vision Alliance. BrainChip’s Akida processor today finds use in a diversity of applications, such as classifying images, identifying odors and tastes, recognizing breath data for disease classification, identifying air quality, interpreting LiDAR laser light data, recognizing keywords, and detecting cybersecurity attacks. Akida leverages advanced neuromorphic computing as its processing “engine”, delivering key features such as one-shot learning and on-device computing with no “cloud” dependencies. As such, it’s particularly valuable in evolving smart “edge” devices, where privacy, security, low power consumption, low latency and high performance, low cost, and small size are all important criteria. In this session, you’ll learn how to easily develop efficient AI in edge devices by implementing Akida IP either into your SoC or as standalone silicon. The presenters will provide detailed performance and other results, derived from real-life system implementations using production Akida silicon, as well as share a variety of design techniques and support resources. For more information and to register, please see the event page. Brian Dipert Editor-In-Chief, Edge AI and Vision Alliance

ASSESSING AND ENHANCING DEEP LEARNING MODELS

Facing Up to Bias Today’s face recognition networks identify white men correctly more often than white women or non-white people. The use of these models can manifest racism, sexism, and other troubling forms of discrimination. There are also publications suggesting that compressed models have greater bias than uncompressed ones. Remarkably, poor statistical reasoning bears as much responsibility for the underlying biases as social pathology does. Further, compression per se is not the source of bias; it just magnifies bias already produced by mainstream training methodologies. By illuminating the sources of statistical bias, as Steve Teig, CEO of Perceive, details in this presentation, we can train models in a more principled way – not just throwing more data at them – to be more discriminating, rather than discriminatory. An Introduction to Data Augmentation Techniques in ML Frameworks Data augmentation is a set of techniques that expand the diversity of data available for training machine learning models by generating new data from existing data. This talk from Rajy Rawther, principal member of the technical staff and software architect at AMD, introduces different types of data augmentation techniques as well as their uses in various training scenarios. Rawther explores some built-in augmentation methods in popular ML frameworks like PyTorch and TensorFlow. She also discusses some tips and tricks that are commonly used to randomly select parameters to avoid having model overfit to a particular dataset.

EMBEDDED AI SOFTWARE DEVELOPMENT TOOLS AND TECHNIQUES

Quickly Measure and Optimize Inference Performance Using Intel DevCloud for the Edge When developing an edge AI solution, DNN inference performance is critical: If your network doesn’t meet your throughput and latency requirements, you’re in trouble. But accurately measuring inference performance on target hardware can be time-consuming—just getting your hands on the target hardware can take weeks or months. In this Over-the-Shoulder tutorial, Corey Heath, Software Engineer at Intel, shows step-by-step how you can quickly and easily benchmark inference performance on a variety of platforms without having to purchase hardware or install software tools. By using Intel DevCloud for the Edge and the Deep Learning Workbench, you can work from anywhere and get instant access to a wide range of Intel hardware platforms and software tools. And, beyond simply measuring performance, using Intel DevCloud for the Edge allows you to quickly identify bottlenecks and optimize your code—all in a cloud-based environment accessible from anywhere. Dynamic Neural Accelerator and MERA Compiler for Low-latency and Energy-efficient Inference at the Edge Achieving high performance and power efficiency for machine learning inference at the edge requires maintaining high chip utilization, even with a batch size of one, while processing high-resolution image data. In this tutorial session, Sakyasingha Dasgupta, the founder and CEO of EdgeCortix, shows how EdgeCortix’s reconfigurable Dynamic Neural Accelerator (DNA) AI processor architecture, coupled with the company’s MERA compiler and software stack, enables developers to seamlessly execute deep neural networks written in Pytorch and TensorFlow Lite while maintaining high chip utilization, power efficiency and low latency regardless of the type of convolution neural network. Dasgupta walks through examples of implementing deep neural networks for vision applications on DNA, starting from standard machine learning frameworks and then benchmarking performance using the built-in simulator as well as FPGA hardware.

UPCOMING INDUSTRY EVENTS

Developing Optimized Systems with BrainChip’s Akida Neuromorphic Processor – BrainChip Webinar: February 24, 2022, 9:00 am PT Embedded Vision Summit: May 17-19, 2022, Santa Clara, California More Events

FEATURED NEWS

BrainChip Achieves Full Commercialization of Its AKD1000 AIoT Chip with Availability of Mini PCIe Boards in High Volume Train Smart Retail AIs 50x Faster with Mindtech’s New Synthetic Data Application Pack e-con Systems Launches 13 Mpixel Monochrome USB 3.1 Gen 1 Camera with High Sensitivity Intel Empowers Developers with oneAPI 2022 Toolkits Syntiant Announces Voice-Enabled Ultra-Low-Power Reference Design for True Wireless Stereo Earbud Applications More News

The post Edge AI and Vision Insights: January 26, 2022 Edition appeared first on Edge AI and Vision Alliance.

This blog post was originally published at Lattice Semiconductor’s website. It is reprinted here with the permission of Lattice Semiconductor.

Sensor fusion is a popular application for Lattice FPGAs, and that particular application is often a topic of discussion among marketing team members here at Lattice. Because of their low power consumption and small size, Lattice FPGAs are often used for sensor fusion in Edge applications. But a recent conversation between the two of us illustrated that what defines an “Edge application” may differ between end markets, and a blog post examining those differences could be informative.

An industrial engineer working in a factory may understand “the Edge” to refer to the boundary between the physical world, in the form of a machine on a production line, and the non-physical world, in the form of the factory’s intranet and/or the internet. In this case, what the engineer considers to be an Edge device may comprise sensors, actuators, and a control system used to monitor the state of a machine and perform any corrective actions as necessary. Oftentimes, such a device may feature some sort of cognitive (reasoning, thinking) capability in the form of artificial intelligence (AI).

Consider the case of machine maintenance, for example, for which there are various strategies that may be employed, including reactive, pre-emptive, and predictive. In the case of reactive maintenance, the idea is to run the machine until it fails and then fix it. The attraction to this strategy is that you can forget about the machine most of the time. The problem is that when the machine does fail, it may disrupt the entire production line, making a potentially minor issue mushroom into a major problem. In the case of pre-emptive maintenance, the machine is serviced, and selected parts are replaced, on a time-based schedule or a number-of-hours-running basis. The advantage of this strategy is that the machine will typically run for a long, long time without problems. The problem is the cost associated with replacing parts before their time and the resources required to perform potentially unnecessary maintenance tasks. When it comes to predictive maintenance, the idea is for the Edge device to use its AI to monitor the machine’s health, looking for anomalies or trends, and then guide the maintenance team to address potential problems before they evolve into real issues.

By comparison, an automotive designer may well regard an entire vehicle as being on the Edge. In this case, the vehicle may boast various advanced driver-assistance systems (ADAS), along with autonomous capabilities and functions, that rely on the gathering of data from myriad sensors, including image (camera), lidar, radar, ultrasonic, and far infrared (FIR) devices. Meanwhile, the task of the various automotive AI systems is to help the driver to drive without hitting anyone or anything, and to protect the vehicle and its human cargo from accidents and mishaps.

A typical complement of automotive sensors used for ADAS and autonomous applications.

In the same way that the term “Edge” can mean different things to different people, so too can the term “sensor fusion.” In its most generic sense, sensor fusion is the process of combining sensory data derived from disparate sources such that the resulting information has less uncertainty than would be possible when these sources were used individually.

One form of sensor fusion involves combining the data from sensors like accelerometers, gyroscopes, and magnetometers to benefit from their strengths and to correct for their weaknesses. In the case of these types of sensors, we might talk about three axes: The X-axis, the Y-axis, and the Z-axis.

The X, Y, and Z axes.

There are two types of motion possible with regard to each of these axes: linear and angular (rotational). In the case of linear motion, movement is possible from side to side along the X-axis, up and down along the Y axis, and forward and backward along the Z axis. These may be regarded as three degrees of freedom (3DOF). By comparison, in the case of angular motion, it’s possible to rotate around one or more of the X, Y, and Z axes, thereby providing another 3DOF. Based on this, we might say that any classical physical system can have a maximum of 6DOF, because there are only six different ways the system can move in three-dimensional (3D) space.

In addition to motion, which we can determine by measuring the differences between the values of all possible DOFs between two times (let’s call them t_n and t_n+1) and then repeating this process over and over again, we may also be interested in orientation (that is, the physical position or direction of the object in question relative to something else), which we can determine by knowing the values of all possible DOFs at some time t_n.

A 3-axis accelerometer measures linear motion along the X, Y, and Z axes. By comparison, a 3-axis gyroscope measures angular rotation around the X, Y, and Z axes. Also, a 3-axis magnetometer can sense where the strongest magnetic force is coming from in the context of the X, Y, and Z axes. Magnetometers are usually used to detect the Earth’s magnetic field, but they can also be used to measure human-made magnetic fields if required.

Each type of sensor has its own strengths and weaknesses. For example, accelerometers are affected by vibration and unaffected by magnetic fields, while magnetometers are unaffected by vibration but may be confused by stray electromagnetic fields. The data from accelerometers can be used to derive rotational information, but gyroscopes provide much more accurate rotational results. On the other hand, gyroscopes are also subject to “drift,” which isn’t an issue with accelerometers and magnetometers.

So, conceptually, the lowest level of sensor fusion involves monitoring the outputs from all three types of sensors and using the data from each pair to correct for errors in the third member of the troika.

The next level of sensor fusion involves combining the data from multiple sensors to provide “situational awareness” that can be used to refine the system’s understanding as to what is occurring in the real world, thereby enabling it to make better decisions. Consider early fitness wearable devices, for example. Although they were reasonably accurate at measuring the number of steps taken on a solid floor, they tended to be confused by exercise machines like treadmills, and a short ride up an escalator might easily end up being counted as an extra 1,000 paces. By comparison, modern equivalents use the combination of sensor fusion and AI to filter out any extraneous noise, to determine if the wearer is walking, running, riding a bicycle, swimming, etc., and to count only legitimate exercise activities.

In the case of automobiles, another type of sensor fusion may be to gather and align-in-time the data from multiple sensors — camera, lidar, radar, etc. — that is presented to AI systems that can compare what the different sensors are reporting and demand extra caution if the conclusions derived from the individual sensors don’t agree.

Not surprisingly, most forms of sensor fusion require the manipulation of large quantities of real-time data with very low latency. Traditional von Neumann processor architectures are not ideally suited for this task. By comparison, field-programmable gate arrays (FPGAs) — such as Lattice CrossLink-NX devices — are ideal for sensor fusion applications because their programmable fabric can be configured to perform sensor processing algorithms in a massively parallel fashion.

Furthermore, CrossLink-NX FPGAs include two hardened 4-lane MIPI D-PHY transceivers running at 10 Gbps per PHY, thereby allowing these devices to provide best-in-class performance for vision processing, sensor fusion, and AI inferencing applications.

The “cream on the top of the cake” is that CrossLink-NX FPGAs are fully supported by the Lattice mVision and sensAI solution stacks. The Lattice mVision solutions stack includes everything embedded vision system designers need to evaluate, develop, and deploy FPGA-based embedded vision applications, such as machine vision, robotics, ADAS, video surveillance, and drones. Meanwhile, the full-featured Lattice sensAI solution stack includes what developers need to evaluate, develop, and deploy FPGA-based AI/ML applications.

Performing sensor fusion (in any of its incarnations) at the Edge (whatever you consider that to be) is advantageous in terms of getting results quickly while conserving communications bandwidth, as opposed to shipping humongous amounts of data into the cloud and then waiting for any results to be returned. Furthermore, the massive parallel processing capabilities offered by low-power, high-performance FPGAs — such as CrossLink-NX devices — means that sensor fusion at the Edge is now both achievable and affordable.

Mark Hoopes
Director of Industrial and Automotive Segment Marketing, Lattice Semiconductor

JP Singh
Automotive Segment Marketing Manager, Lattice Semiconductor

The post Sensor Fusion at the Edge: Two Perspectives appeared first on Edge AI and Vision Alliance.

25 January 2022 – Jerusalem – Two Israeli AI startups, Visionary.ai and Inuitive, today announced a new partnership. The companies have been collaborating in recent months to run Visionary.ai’s AI-based imaging technology on Inuitive’s AI processor, making cutting-edge computer vision available to edge devices, affordably and with minimal power consumption.

Visionary.ai has developed a software-based image signal processor (ISP) which leverages AI to help cameras achieve market-leading image quality and detection accuracy. Inuitive, a semiconductor company, develops and produces computer vision processors with AI accelerators.

The two companies will start showcasing their joint demo to prospective customers later this month. They plan to demonstrate Inuitive’s market-leading AI processor with Visionary.ai’s software-based ISP inside.

Today, standard ISPs discard a critical portion of sensor data. Visionary.ai’s ISP leverages all the raw sensor data and cutting-edge AI, presenting a challenge of high compute power.

Commenting on the partnership, Visionary.ai co-founder and CEO Oren Debbi said, “Running on the Inuitive processor means we can offer stellar image quality to so many more camera and electronics manufacturers without them paying the price of high compute power.

“With this new advantage, and having a partner like Inuitive on board, will help us unlock a huge market opportunity. In the coming years, we expect to see the package of our two game-changing technologies being adopted globally”.

Shlomo Gadot, founder and CEO at Inuitive commented: “Partnering with Visionary.ai offers our customers the most advanced vision and AI solutions.

“Leveraging the industry leading Edge AI processor series from Inuitive and Visionary.ai’s sophisticated AI-based ISP will enable vision systems to operate at a very wide dynamic range of use cases, with the highest levels of performance while keeping system’s simplicity.”

Reflecting on their first year of R&D, Visionary.ai co-founder and CTO Yoav Taieb added, “This milestone is a meaningful development to reach at the end of our first 12 months in business.

“During our inception phase, we were running on a server, and then as we reduced our compute power, moved to the Nvidia AGX. With Inuitive, we’re now able to run in real-time with a high frame rate on a low-cost edge processor.”

The post Real-time Video Enhancement Reaches New Standard with Visionary.ai and Inuitive Partnership appeared first on Edge AI and Vision Alliance.

This blog post was originally published at e-con Systems’ website. It is reprinted here with the permission of e-con Systems.

Over the past few years, embedded vision has evolved from a buzzword to a widely adopted technology used across industrial, medical, retail, entertainment, and farming sectors. With each phase of its evolution, embedded vision has ensured a significant growth in the number of camera interfaces available to choose from. However, despite the technological advancements, MIPI and USB interfaces have remained the two most popular types for a majority of embedded vision applications.

The selection of a best-fit interface depends on many factors like frame rate/bandwidth requirements, resolution, data transfer reliability, cable length, complexity, and – of course – the overall cost. In this article, we look at both interfaces in detail to better understand their capabilities and limitations.

A deeper look at MIPI and USB interfaces

A MIPI camera is nothing but a camera module or system that uses a MIPI interface to transfer images from the camera to the host platform. In comparison, a USB camera uses a USB interface for data transfer. Now, let us understand the different types of MIPI and USB interfaces and where they are used.

MIPI Interface

MIPI is the most commonly used interface in today’s market for point-to-point image and video transmission between cameras and host devices. It can be attributed to MIPI’s ease of use and its ability to support a broad range of high-performance applications. It also comes equipped with powerful features such as 1080p, 4K, 8K and beyond video and high-resolution imaging.

MIPI interface is an ideal choice for applications like head-mounted virtual reality devices, smart traffic applications, gesture recognition systems, drones, facial recognition, security, surveillance systems, etc.

MIPI CSI-2 Interface

The MIPI CSI-2 (MIPI Camera Serial Interface 2nd Generation) standard is a high-performance, cost-effective, and simple-to-use interface. MIPI CSI-2 offers a maximum bandwidth of 10 Gb/s with four image data lanes – each lane capable of transferring data up to 2.5 Gb/s. MIPI CSI-2 is faster than USB 3.0 and has a reliable protocol to handle video from 1080p to 8K and beyond. In addition, owing to its low overhead, MIPI CSI-2 has a higher net image bandwidth.

The MIPI CSI-2 interface uses fewer resources from the CPU – thanks to its multi-core processors. It is the default camera interface for Raspberry Pi and Jetson Nano. The Raspberry Pi camera module V1 and V2 are also based on it.

Limitations of MIPI CSI-2 Interface

Even though it is a powerful and popular interface, MIPI CSI does come with a few limitations. For instance, MIPI cameras rely on extra drivers to work. It means that there is limited support for different image sensors unless embedded system manufacturers really push for it!

To learn more about MIPI cameras and how they work, have a look at What is a MIPI camera? How does MIPI camera work?

USB Interface

The USB interface tends to serve as the junction between two systems – the camera and the PC. Since it is well-known for its plug-and-play capabilities, choosing the USB interface implies that you can say goodbye to expensive, drawn-out development times and costs for your embedded vision interface. USB 2.0, the older version, has significant technical limitations. As the technology begins to dwindle, a number of its components become incompatible. USB 3.0 and the USB 3.1 Gen 1 interfaces were launched to overcome the limitations of the USB 2.0 Interface.

USB 3.0 Interface

The USB 3.0 (and USB 3.1 Gen 1) interface combines the positive features of different interfaces. These include plug-and-play compatibility and low CPU load. The vision industrial standard of USB 3.0 also increases its reliability for high-resolution and high-speed cameras.

It requires minimal additional hardware and supports low bandwidth – up to 40 megabytes per second. It has a maximum bandwidth of 480 megabytes per second. This is 10 times faster than USB 2.0 and 4 times faster than GigE! Its plug-and-play capabilities ensure that embedded vision devices can be swapped out with ease – making it easy to replace a damaged camera.

Limitations of USB 3.0 Interface

The biggest disadvantage of the USB 3.0 interface is that you cannot run high-resolution sensors at high speed. Another downfall is that you can only use a cable up to a distance of 5 meters from the host processor. While longer cables are available, they are all fitted with “boosters”. How well these cables work together with industrial cameras has to be checked for every individual case.

MIPI Camera vs USB Camera – a feature by feature comparison

Now that we understand what each of the two interfaces is, let us do a feature by feature comparison of MIPI and USB in the below table:

• – Max bandwidth with MIPI-CSI2 / MIPI D-PHY® v1.2 used in NVIDIA Jetson AGX Xavier Platform

GMSL is another popular interface which is used as an alternative to MIPI and USB, typically when the image data has to be transferred to a distance of more than 3 meters. To understand how a MIPI camera compares with GMSL camera, check out GMSL camera over MIPI camera.

To learn how to choose an interface for your embedded vision system, visit How to choose the right interface for an embedded vision system.

Related Cameras

• MIPI CSI-2 Camera Modules
• See3CAM – USB 3.0 Cameras
• USB 2.0 Cameras

Conclusion

A camera interface represents a make or break moment for embedded vision applications. Hence, picking the right interface between MIPI and USB is a crucial decision. Even though there are some embedded vision applications where neither MIPI nor USB will be a fit, in majority of the cases one of them will suffice. Choosing the best fit interface will depend on the feature set required by your embedded device.

Prabu Kumar
Chief Technology Officer and Head of Camera Products, e-con Systems

The post MIPI Cameras vs USB Cameras: a Detailed Comparison appeared first on Edge AI and Vision Alliance.

Immersive extended reality (XR) experiences let users seamlessly interact with virtual environments. These experiences require real-time gesture control and eye tracking while running in resource-constrained environments such as on head-mounted displays (HMDs) and smart glasses.

These capabilities are typically implemented using computer vision technology, with imaging sensors that generate lots of data to be moved and processed without any selectivity, costing both latency and power. Most of this data could be discarded at the edge without any loss of useful information, just like Oculi’s technology does.

Oculi helps to solve low sensor speed and high external processing limitations by enabling near-zero lag performance with an embedded solution for eye or gesture tracking as shown in this video. Note that the webcam is inverted so the hand movements are mirrored as compared to the grid output.

Get in touch with us at info@oculi.ai if you would like to see more of our demos!

The post Oculi Enables Near-zero Lag Performance with an Embedded Solution for Gesture Control appeared first on Edge AI and Vision Alliance.

On March 2-3, 2022, Alliance Member company FRAMOS will deliver the online training “Characterization of a CMOS Camera”. From the event page:

Everyone who is interested in understanding how to obtain the performance characteristics of a CMOS image sensor/camera and how to communicate the results to their peers and customers. The training is accessible to designers, development, and test engineers that deal with CMOS image sensors and CIS cameras. The training is primarily an in-depth technical class. Because the training is very technical, participants need to have a technical background, preferably in the field of digital imaging.

Top key takeaways from this training are:

1. Get familiar with the various performance parameters of a CMOS camera
2. Learn how to measure the characteristics of a CMOS camera
3. Reading data sheets of a CMOS camera
4. Get familiar with the pitfalls of obtaining the right figures and numbers of the CMOS camera performance

For more information and to register, please see the event page.

The post Upcoming Online Training Explores CMOS Camera Characterization appeared first on Edge AI and Vision Alliance.

MONTREAL, QC – January 18, 2022 – Algolux, a globally recognized computer vision software company, today announced that it has been named a CES® 2022 Innovation Awards Honoree for its Eos Embedded Perception Software, delivering industry-leading robustness for automotive advanced driver assistance systems (ADAS) and autonomous vehicles. This year’s CES Innovation Awards program received a record high number of over 1800 submissions. The announcement was made following CES 2022, the world’s most influential technology event, held Jan. 5-8 in Las Vegas, NV and digitally.

Automotive ADAS and autonomous vehicle systems are the key to improving road safety and unlocking the future of transportation via new mobility services, such as self-driving shuttles and automated trucking. But these systems fail due to a lack of computer vision robustness in common but challenging driving scenarios, such as poor weather conditions and low light.

Eos is an award-winning embedded perception solution that massively improves vision system robustness in all conditions while being easily scaled to any vision system architecture. Benchmarked by many of the top OEMs and Tier 1 suppliers in the automotive and autonomous vehicle industries, Eos has been shown to outperform the leading commercial alternatives by up to 3x in detection accuracy, especially in these difficult scenarios.

The CES Innovation Awards program, owned and produced by the Consumer Technology Association (CTA)®, is an annual competition honoring outstanding design and engineering in 27 consumer technology product categories. Those with the highest rating receive the “Best of Innovation” distinction. An elite panel of industry expert judges, including members of the media, designers, engineers and more, reviewed submissions based on innovation, engineering and functionality, aesthetic and design.

“Robust perception in all conditions is critical for the safe operation of consumer vehicles, commercial transportation, and various forms of autonomy,” said Allan Benchetrit, Algolux CEO. “Algolux is thrilled to have our Eos Embedded Perception Software platform recognized by the CES 2022 Innovation Awards panel of global experts and influential media for its impact on safety in automotive ADAS and autonomous vehicles.”

The CES 2022 Innovation Awards honorees, including product descriptions and photos, can be found at CES.tech/innovation.

Owned and produced by CTA, CES 2022, the global stage for innovation, convened the tech industry in person and digitally, giving global audiences access to major brands and startups, as well as the world’s most-influential leaders and industry advocates.  Visit CES.tech for all CES 2022 updates.

About Algolux

Algolux is a globally recognized computer vision company addressing the critical issue of safety for advanced driver assistance systems and autonomous vehicles. Our machine-learning tools and embedded AI software products enable existing and new camera designs to achieve industry-leading performance across all driving conditions. Founded on groundbreaking research at the intersection of deep learning, computer vision, and computational imaging, Algolux has been recognized at numerous industry and academic conferences and has been named to the 2021 CB Insights AI 100 List of the world’s most innovative artificial intelligence startups.

Algolux is headquartered in Montreal, Canada, with offices in Palo Alto, California, and Munich, Germany. Learn more at algolux.com.

The post Algolux Named as CES 2022 Innovation Awards Honoree for Automotive ADAS and Autonomous Vehicles Solutions appeared first on Edge AI and Vision Alliance.

The PDF file linked to below is the presentation from the January 20, 2022 Edge AI and Vision Alliance Member Briefing. Please be aware that these materials are for Alliance Member company internal use only. January 20, 2022 Edge AI and Vision Alliance Member Briefing  

January 20, 2022 Edge AI and Vision Alliance Member Briefing Presentation

Register or sign in to access this content.

Registration is free and takes less than one minute. Click here to register and get full access to the Edge AI and Vision Alliance's valuable content.

The post January 20, 2022 Edge AI and Vision Alliance Member Briefing Presentation appeared first on Edge AI and Vision Alliance.

In this video, Robert Lundberg, CTO and Co-founder of PerceptiLabs, builds a +80% accuracy model that detects anomalies in magnetic tiles in under 5 minutes!

• Try out PerceptiLabs (free for developers!): https://www.perceptilabs.com
• Dataset used in the video: https://www.kaggle.com/alex000kim/magnetic-tile-surface-defects/version/1
• Read our blog: https://blog.perceptilabs.com
• Leave your questions and feedback: https://forum.perceptilabs.com
• Join our friendly community: https://perceptilabs-com.slack.com/join/shared_invite/zt-mxohtpkm-lQI_0nT~tBeUd2RtZIfe4g#/shared-invite/email

The post Instant Models: Detecting Anomalies in Magnetic Tiles appeared first on Edge AI and Vision Alliance.

This blog post was originally published at ClearML’s website. It is reprinted here with the permission of ClearML.

The infamous data science workflow with interconnected circles of data acquisition, wrangling, analysis, and reporting understates the multi-connectivity and non-linearity of these components. The same is true for machine learning and deep learning workflows. I understand the need for oversimplification is expedient in presentations and executive summaries. However, it may paint unrealistic pictures, hide the intricacies of ML development and conceal the realities of the mess. This brings me to the tools of the trade or more commonly referred as the infrastructure of artificial intelligence which is the vehicle under which all libraries, experimentations, designs and creative minds meet. These humble infrastructures tend to be overlooked and underappreciated but their glaring importance can’t be overstated. I will explain a couple of these very tools that could be used in tandem to improve your workflow, accountability of data exploration, and lower the time and resources to go from proof-of-concept (POC) to deployment.

Simplified ML/DL Pipeline

Open-Source Software

Before you are on the stage with a turtleneck shirt and casual blue jeans to reveal your data-intensive breakthrough product, the first step is to choose software to tackle the complex terrains of machine learning. Lucky for you, there are excellent open-software platforms such as Apache Spark, Jupyter, TensorFlow, PostgreSQL, and many more to get you started. The recent addition to my list is Allegro Trains which I am a huge proponent of. As a working data scientist, I have five criteria before choosing to use open-source software. These are:

1. Availability and ease of use for POC.
2. Reproducibility and collaboration.
3. Cross-platform compatibility.
4. Resource optimization.
5. Productivity boost relative to the learning rate.

In the spirits of these criteria, let’s explore Jupyter and Trains.

Jupyter and Allegro Trains

Jupyter notebook is the de facto gateway to interactive computing for aspiring data scientists and engineers. It offers quick feedback in the form of errors or visual displays and allows users to test out their rudimentary ideas. It’s undoubtedly the perfect tool for POC in addition to being language agnostic and offering text through Markdown cells. Jupyter Lab adds modularity to the notebooks so you are not opening multiple tabs with each notebook. There is also JupyterHub which allows multiple users to share notebooks. These are all great platforms to get some hands-on experience.

The problem comes when you are tackling complex projects where a large number of iterations and experimentations, crucial code and resource optimizations, versioning and collaboration, and data integration and deployment are needed. These are the bottlenecks of bringing data-driven products to the market. As a result, AI architectures that incorporate these features in addition to the five criteria above become a necessity to accelerate R&D to market stages with faster ROI.

Machine learning project management by Trains

Why Allegro Trains?

Take a look at the dummy four projects in figure 1 and imagine if you were running these projects on Jupyter, the level of difficulty to keep the purpose of each notebook quickly becomes a task of its own. You could solve this issue by setting up a git repository to handle your versioning needs and manage your progress. You could also install a memory profiler to manage your resource allocation to each task. What about automating your ML models? You could explore the growing list of AutoML. If you are starting to realize that your ML project complexity is rising with each step to completion, then you are not the only one. This is a common problem and the solution should be addressed with a singular project hub. This is precisely why Allegro Trains was developed.

Allegro Trains puts a stop to these all-too-common infrastructure problems. Instead of patching your problems with countless modules, it’s time to bring maturity to ML with an encompassing solution. Trains provides a simple MLOps and AI infrastructure solution to speed up machine/deep learning projects from R&D to deployment. It allows data professionals to take charge of implementing creative insights without the need to worry about messy notebooks, interruption to solve versioning issues, model comparisons, automatic logging, performance tracking, and even CPU/GPU/IO resource allocations. In short, it meets my five-criterion for using open-source software. It also has AutoML to expedite experimentation and the capability to terminate resource-intensive or underperforming tasks from the Web UI.

I would like to leave you with this last remark. The ability to use open-platforms in tandem to bring your exciting ideas and R&D to market has become a matter of simple installations. Jupyter is there for your needs of exploration and ClearML tracks your every progress from POC to optimization and collaboration across teams to deployment with Trains Server.

Henok Yemam
Data Scientist, FINRA

The post Machine Learning with Jupyter: Solving the Workflow Management Problem Using Open Platforms appeared first on Edge AI and Vision Alliance.

Munich, January 19, 2022 – FRAMOS, a global leader in machine vision systems, is well placed to respond to rising demand in the vision market despite upstream product shortages (e.g. semiconductors), or pandemic-related supply chain delays. FRAMOS expects to continue to fulfill customer orders on time and in budget during these difficult times and throughout 2022. FRAMOS is helping its customers achieve their development goals while meeting their time-to-market and cost targets with its managed supply of key products.

More Than Just Semiconductors In Short Supply

These are challenging times for the industry: global supply chains are strained by compounding material shortages, for example, the general scarcity of semiconductors and a lack of microchips due to unprecedented demand and growth in IoT and 5G (e.g. smartphones). On top of that, carmakers are switching to electric vehicles and, of course, the global COVID-19 pandemic have brought further challenges for logistics besides worker shortages. This has had a trickledown effect that was even felt by retailers having to cope with shortages during the past Christmas season. An assessment by FRAMOS indicates that the vision industry will not see any noticeable improvement during 2022.

Impacts on Embedded Vision Component Availability

A lack of key upstream products, such as semiconductors, will make things difficult for many industries. Having initially impacted the automotive sector, ongoing supply difficulties and rising costs are now affecting the vision industry too – products that are in particularly high demand like lenses manufactured in mainland China and rolling shutter image sensors in limited supply. Scarcity in the semiconductor market may lead to shortages of microcontrollers (MCUs), microprocessors (MPUs), graphics processors (GPUs) and FPGAs (integrated circuits) adding further strain to manufacturers. All these components play an essential role in the development and production of embedded vision systems and cameras.

FRAMOS Mitigation Plans

The vision market is expanding and offers good growth opportunities, so it is vital to maintain the ability to deliver components and products. Ron Low, VP Partners and Channels at FRAMOS, says: “FRAMOS has optimized warehouse inventory management systems and made preparations early on pre-pandemic to maintain optimum service levels to customers. Current strategies extend well beyond warehousing and include proactive forward planning, a well thought out and running procurement methodology coupled with many years of experience and feel for developments in the vision market. As a result, the company is in an excellent position to respond to the existing challenges and minimize procurement risk for those more particularly affected product areas”.

In recent years, both Sony and Intel® have worked closely with FRAMOS worldwide to meet customers’ ongoing need for vision components during the pandemic. The vast majority of customers have not experienced any delivery delays thanks to strong partners and in-house efforts. They were comforted in knowing that they could rely on FRAMOS to uphold its service levels even when the supply situation, component availability and demand requirements changed considerably.

Delivery Reliability – The Foundation for Trust and Cooperation

FRAMOS’s stable product supply, excellent communication and good reputation in the vision market led many new customers to partner with FRAMOS in 2021. These relationships enabled customers to meet their development schedules and time-to-market targets while securing market share. Preserving our delivery capability on stable terms for our own products in the FRAMOS Sensor Module Ecosystem and for imaging components from an international network of well-known partners is immensely important to our customers. It is also one of the keys that will lead to further growth and success for FRAMOS in 2022 and beyond.

About FRAMOS

FRAMOS® is an imaging expert, trusted advisor, and vision solutions provider. Since 1981, FRAMOS implements the best current and emerging imaging technologies to address specific customer requirements and applications. FRAMOS meets these requirements with advanced and proven imaging components from a global network of renowned partners and with FRAMOS IP.

More than 180 FRAMOS employees world-wide are passionate about the unlimited potential of imaging and help customers achieve the optimum results from every possible scenario. FRAMOS drives and ensures the entire product development journey from POC, through prototyping, to mass production. FRAMOS carefully selects imaging components, like image sensors, lenses, or various 3D technologies, and offers custom developments tailored to individual needs and time frames. FRAMOS listens and understands customer challenges. With innovative solutions FRAMOS ensures successful project outcomes and develops long-term customer relations.

For more information visit: www.framos.com, Twitter or LinkedIn.

The post FRAMOS: Delivery Assurances, Price Stability Through Strong Partners appeared first on Edge AI and Vision Alliance.

Strong industry consensus to develop an open, cross-vendor API standard for portable control over camera systems in multiple markets

Beaverton, OR – January 19, 2022 – 6:00 AM PT – Today, The Khronos® Group, an open consortium of industry-leading companies creating advanced interoperability standards, together with the European Machine Vision Association (EMVA) announces the formation of a new Khronos Working Group to develop an open, royalty-free API standard for controlling camera system runtimes in embedded, mobile, industrial, XR, automotive, and scientific markets. Over 70 companies participated in an EMVA/Khronos-hosted Exploratory Group during 2021 to develop a Scope of Work document that will guide the direction of the API design. Design work is expected to start in February 2022, and any organization is invited to join Khronos to participate.

Cameras are increasingly critical in diverse markets, motivating the development of increasingly sophisticated optical systems, image sensors, and vision processors often utilizing machine learning technology. However, the lack of interoperable camera API standards increases application development time and maintenance costs while reducing portability and opportunity for code reuse, resulting in unnecessarily high integration costs for camera technologies.

“The Embedded Camera API Exploratory Group followed the Khronos New Initiative Process with invaluable cooperation from the EMVA. Over seventy companies worked together from March to December 2021 to forge strong industry consensus on the need, terminology, scope, requirements, and design methodology for a new open standard camera system API,” said Neil Trevett, Khronos president. “Now, we warmly invite any interested companies, vendors and developers to bring their voice and their expertise to the design phase of this important work.”

The Camera API will be designed to provide applications, libraries and frameworks low-level, explicit control over camera runtimes, with a low-level of abstraction that still provides application portability over a wide variety of camera systems with effective, performant control to generate streams of data for consumption by downstream applications and clients.

“The close and productive collaboration between the EMVA and Khronos has been very effective in enabling a broader industry participation and diversity of perspectives at the Embedded Camera Exploratory Group than either organization could have achieved working alone,” said Chris Yates, EMVA president. “EMVA will continue to work closely with Khronos under our new liaison agreement to ensure that the interests of both the EMVA membership and the wider industry are represented at the new Camera API working group.”

The Camera API Working Group will start meetings in February 2022 and is expected to be of particular interest to sensor or camera manufacturers, silicon vendors, and software developers working on vision and sensor processing. Any organization is welcome to join Khronos and participate in this global initiative under the consortium’s multi-company governance process. Please visit the Khronos membership page or contact memberservices@khronos.org for more details.

Industry Support for the Camera API Working

Over 70 companies participated in the Camera Exploratory group and the following companies support establishing the Camera API Working Group: Adimec, Almalence Inc., Analog Devices Inc., Basler AG, Baumer Optronic GmbH, Cadence Design Systems, Inc., Collabora, Digica, Digital Air Technologies, Euresys, European Machine Vision Association, FLIR Integrated Imaging Solutions, Google, Groget, Holochip Corporation, Ideas on Board Oy, LunarG Inc., MATRIX VISION, MM Solutions, MVTec Software GmbH, NVIDIA, Perey Research & Consulting, Phil-Vision, Pleora Technologies, Raspberry PI Ltd, STEMMER IMAGING, Texas Instruments, VeriSilicon, Vision Components.

“The generic camera API will help Adimec to focus on our mission to deliver the right image in the right place at the right time, so our customers can focus on their imaging tasks. That is what we call ‘Excellence in Imaging’,” – The Adimec Team.

“Lack of API standards for advanced use of embedded cameras and sensors is an impediment to industry growth, collaboration and innovation. Enterprise AR customers and systems integrators/value added providers will benefit from greater clarity, open interfaces between modular systems and innovation in the component provider ecosystem. This Khronos standard for camera and sensor control will increase opportunities for powerful new combinations of sensor and AR compute resources, integration with existing IT, and lower cost and complexity of future solutions,” Christine Perey, interoperability and standards program leader for the Augmented Reality for Enterprise Alliance (AREA).

“Open interface standards such as GenICam or GigE Vision have been a key element to establish a professional Machine Vision Market. Only by such standards we can ensure the interoperability of products from different vendors. I​t helped to shorten the development cycles of customers dramatically and also yields in a faster growing market. Therefore we strongly support the new open standard camera API initiative driven by Khronos and the EMVA,” Arndt Bake, CDO, Basler AG.

“Over the past two decades, digital cameras used in embedded applications have changed dramatically. As video capture quality and processing power have increased, so has the potential for enhanced features which were unimaginable in early camera phones. The proliferation of features has resulted in a corresponding plethora of software support. The Embedded Camera Exploratory Group has laid the foundations for a consistent and extensible API to resolve this complexity; Digica is pleased to have contributed to this project and welcomes the development of the API under the new Working Group,” Jim Carroll, CTO, Digica.

“Due to high fragmentation and lack of standardization, the embedded camera space is subject to painful interoperability issues. Adding camera support in a product is complex and expensive, most often subject to vendor lock-in, when not practically impossible for small actors. Ideas on Board launched the libcamera project three years ago to address these issues in the Linux mobile, embedded and desktop ecosystems. We have contributed our experience to the Khronos Camera Exploratory Group, and are looking forward to continuing collaboration with the industry on a new open standard camera API,” said Laurent Pinchart, CEO, Ideas on Board, and lead architect of the libcamera® project.

“Cameras are everywhere and in everything, the market and applications have exploded in the last ten years. But a cohesive set of standard APIs has been slow to emerge making incompatibility challenging. Khronos, in conjunction with the European Machine Vision Association, is going to correct that and has formed this Working Group to develop an open API for cameras. This will be welcome news to industry participants and users alike,” said Jon Peddie, president, Jon Peddie Research.

“Existing standards, like GigE Vision and USB3 Vision, have proven that a standardization of software interfaces is beneficial for manufacturers and users. We believe that, in the rapidly changing world, Embedded Vision is significantly shaping the future of machine vision. A complementary standard for the embedded camera API is therefore important, and it makes camera control more reliable, hardware selection more flexible and shortens users’ time-to-market,” said Tilman Sanitz, head of embedded systems, Matrix Vision.

“A widely supported open standard camera API will spur innovation and reduce integration costs in multiple markets that use advanced sensors. NVIDIA has supported the work of the Exploratory Group and is committed to participating in the design work at this new Camera Working Group,” Sean Pieper, director of imaging software, NVIDIA.

“With the strong growth of camera applications in automotive, IoT, AR/VR devices, wearables, and smartphones, there has been a strong demand for a standardized camera API in the industry. The standardized camera API that the Khronos group is working on will help facilitate the deployment of new cameras by reducing porting efforts, simplifying the procedures of camera upgrades, and improving the interoperability among various camera devices. This camera API standardization effort is very meaningful and will be highly influential to the related industry. We would like to see this standard API to be deployed soon,” said Weijin Dai, EVP, VeriSilicon.

About EMVA

The European Machine Vision Association (EMVA) is a non-for-profit and non-commercial association representing the Machine Vision industry in Europe that is open for all types of organizations having a stake in machine vision, computer vision, embedded vision or imaging technologies: manufacturers, system and machine builders, integrators, distributors, consultancies, research organizations and academia. The EMVA hosts four international vision standards, and all members – as the 100% owners of the association – benefit from the dedicated networking, standardization, and cooperation activities of the EMVA.

About Khronos

The Khronos Group is an open, non-profit, member-driven consortium of over 180 industry-leading companies creating advanced, royalty-free, interoperability standards for 3D graphics, augmented and virtual reality, parallel programming, vision acceleration and machine learning. Khronos activities include 3D Commerce, ANARI, glTF, NNEF, OpenCL, OpenGL®, OpenGL® ES, OpenVG, OpenVX, OpenXR, SPIR-V, SYCL, Vulkan®, and WebGL. Khronos members drive the development and evolution of Khronos specifications and are able to accelerate the delivery of cutting-edge platforms and applications through early access to specification drafts and conformance tests.

The post Khronos and EMVA Issue Call for Participation for New Camera API Working Group appeared first on Edge AI and Vision Alliance.

In this video, you’ll see how AI technology can be used to identify faces using the MAX78000EVKIT from Maxim Integrated (now part of Analog Devices), along with a camera and display. This face identification demonstration, shown by Erman Okman and Gorken Ulkar, members of the technical staff at Maxim Integrated, implements an energy-efficient convolutional neural network inference engine trained by the MAX78000 SDK, which supports PyTorch and TensorFlow.

The post AI Face Identification Using the Maxim Integrated (Analog Devices) MAX78000EVKIT appeared first on Edge AI and Vision Alliance.

This blog post was originally published at Opteran Technologies’ website. It is reprinted here with the permission of Opteran Technologies.

When I interviewed at Beats by Dr. Dre in 2012, I had a long conversation with Jimmy Iovine about how he and Dr. Dre founded Beats. He started by teaching me that a music producer’s job is to create the best version of a song by bringing out the greatness in the artist. He regaled me with stories of monomania for getting the exact right snare drum sound, flexing with the unpredictable human element of the artist, and obsessive tinkering at the sound mixing board, turning knobs and moving sliders, over and over again.

I learned that only the music producer and the artist know how a song truly sounds. As Jimmy explained, they are the only people who heard the final version of the song playback in that music studio, at that moment in time, when they both agreed the song was done. After that, the song goes scattershot out into the world, playing over car radios and through crackly speakers.

With the advent of devices like music players, mobile phones, and iPods, access to music increased but the quality of the audio decreased due to the cheap in-the-box earphones that accompanied these devices. A listener could hear the music, but it was not the song the producer or artist intended; the audio was tinny, highs and lows were dropped, and it was often muddy.

The solution for this problem already existed in 2009 when Beats launched: highly engineered, expensive, large over-the-ear headphones, the equipment that all producers and artists wore in the music studio. Those studio headphones were specifically designed for professional audio quality and thus, the product was limited to the music industry and wealthy audiophiles. No one seriously considered that pre-existing solution for the mass market.

When Jimmy floated the idea of Beats headphones for consumers to his network, many only saw obstacles: kids will never wear big, clunky headphones, average music consumers will never pay hundreds of dollars for headphones, and the technology and engineering are too complex to become affordable for the mass market. By asking a First Principles question – what currently delivers great audio – Jimmy and Dre had the answer, it was already on their ears. To make it an opportunity, they had to add youth culture appeal and value engineering. The $3B sale to Apple in 2014 proved they got it right.

Opteran

Opteran is doing the exact same First Principles, pattern matching that Jimmy and Dre did back in 2009. The problem is not lousy audio; it is the massive challenge of delivering on the promise of autonomous machines and vehicles to create better jobs, better safety, and better communities. Autonomous machines monitoring our health and delivering life-saving medication and assistance; autonomous vehicles rescuing and transporting humans safely away from natural disasters; autonomous industrial machines working routinely on inhospitable environments; and the lists go on.

Autonomous machines and vehicles need to See, Sense, Direct and Decide how and what to do, at all moments in time. Fail-safe software interacting with complex hardware elements must be built for each of these elements. For many years now, efforts have focused on replicating the “human” brain through Machine Learning and the job is getting done with tradeoffs: brittle and opaque neural networks, voracious energy consumption, and unending data feeding and teaching of unlimited changing conditions.

Opteran queried what else besides humans can See, Sense, Direct and Decide and just like with Beats, the answer already existed. Nature has evolved perception, navigation and decision-making for hundreds of millions of years. The Natural Intelligence of insect brains are Opteran’s music studio headphones. The one million neurons of the honeybee brain can perform complicated actions including flying and landing, foraging, and walking around the complex, changing structures of flowers, plants and honeycombs. Opteran brings this natural autonomy to silicon, including being resilient, low power and light weight. Based on over eight years of University research, the Opteran Development Kit (ODK) is the start to enabling machines and vehicles to move just like movement in nature.

Joining the Opteran Board

As the Non-Executive Director to Opteran, my job is to bring my independent judgment to bear on strategy, performance, key appointments and standards of conduct. I was lucky to join Beats and to help build it out globally. Today, I feel just as fortunate to join the Opteran Board of Directors to use my experience and network to help Opteran bring Natural Intelligence to the market. Having helped scale a company built on First Principles at Beats, Opteran’s endeavor to solve today’s movement-based autonomy by looking to nature, where these problems are already solved, is music to my ears.

Rafferty Jackson
Non-Executive Director, Opteran Technologies

The post What Do Headphones and Insect Brains Have in Common? appeared first on Edge AI and Vision Alliance.

Company welcomes enterprises to the Edge with AIoT PCIe board and board design layout

Laguna Hills , Calif. – DATE, 2022 – BrainChip Holdings Ltd (ASX: BRN, OTCQX: BRCHF, ADR: BCHPY), a leading provider of ultra-low power high performance artificial intelligence technology and the world’s first commercial producer of neuromorphic AI chips and IP, today announced that it has begun taking orders for the first commercially available Mini PCIe board leveraging its Akida advanced neural networking processor, rounding out its suite of AKD1000 offerings.

The AKD1000-powered Mini PCIe boards can be plugged into a developer’s existing system to unlock capabilities for a wide array of edge AI applications, including Smart Home, Smart Health, Smart City and Smart Transportation. BrainChip will also offer the full PCIe design layout files and the bill of materials (BOM) to system integrators and developers to enable them to build their own boards and implement AKD1000 chips in volume as a stand-alone embedded accelerator or as a co-processor.

The new boards help usher in a new era of AI at the edge due to their performance, security, low power requirements, and the ability to perform AI training and learning on the device itself, without dependency on the cloud. The production-ready chips provide high-speed neuromorphic processing of sensor data at a low cost, high speed and very low power consumption.

The PCIe boards are immediately available for pre-order on the BrainChip website. Pricing starts at $499.

“I am excited that people will finally be able to enjoy a world where AI meets the Internet of Things,” said Sean Hehir, BrainChip CEO. “We have been working on developing our Akida technology for more than a decade and with the full commercial availability of our AKD1000, we are ready to fully execute on our vision.  Other technologies are simply not capable of the autonomous, incremental learning at ultra-low power consumption that BrainChip’s solutions can provide. Getting these chips into as many hands as possible is how the next generation of AI becomes reality.”

The launch of BrainChip’s new PCIe board closely follows the company’s development kit offerings introduced in October.  The two development kits – an x86 Shuttle PC development kit, as well as an ARM-based Raspberry Pi development kit – both include the AKD1000 chip on a Mini PCIe board and are available to partners, large enterprises and OEMs. BrainChip’s AKD1000 chips and PCIe board can be purchased at shop.brainchipinc.com or via the Buy Now button at www.brainchip.com/.

About BrainChip Holdings Ltd (ASX: BRN, OTCQX: BRCHF, ADR: BCHPY)

BrainChip is a global technology company that is producing a groundbreaking neuromorphic processor that brings artificial intelligence to the edge in a way that is beyond the capabilities of other products. The chip is high performance, small, ultra-low power and enables a wide array of edge capabilities that include on-chip training, learning and inference. The event-based neural network processor is inspired by the spiking nature of the human brain and is implemented in an industry standard digital process. By mimicking brain processing BrainChip has pioneered a processing architecture, called Akida, which is both scalable and flexible to address the requirements in edge devices. At the edge, sensor inputs are analyzed at the point of acquisition rather than through transmission via the cloud to a data center. Akida is designed to provide a complete ultra-low power and fast AI Edge Network for vision, audio, olfactory and smart transducer applications. The reduction in system latency provides faster response and a more power efficient system that can reduce the large carbon footprint of data centers.

Additional information is available at https://www.brainchipinc.com

Follow BrainChip on Twitter: https://www.twitter.com/BrainChip_inc

Follow BrainChip on LinkedIn: https://www.linkedin.com/company/7792006

The post BrainChip Achieves Full Commercialization of Its AKD1000 AIoT Chip with Availability of Mini PCIe Boards in High Volume appeared first on Edge AI and Vision Alliance.

• Smart Retail application pack enables developers to rapidly train AIs with synthetic data for next-gen use-cases

• Current retail customer use cases include enhancing customer safety, spill and object detection, monitoring stock levels, customer in-store engagement and traffic flow

• Mindtech will be exhibiting at NRF on booth # 1064

JANUARY 17, 2022 – NEW YORK – Mindtech Global, developer of the world’s leading synthetic data creation platform for training AI vision systems – has released a new application pack for companies developing smart retail systems.

The retail application pack is now available on Mindtech’s Chameleon platform. The platform allows developers to create unlimited amounts of synthetic training data to help computers see, understand and predict human-to-human and human-to-world interactions.

Chameleon’s new retail application pack has pre-built 3D retail assets such as retail environments and car parks, together with automated behaviors relevant to retail use cases. Using the Chameleon platform, customers developing AI vision systems for retail can quickly build unlimited scenes and scenarios using the application pack’s photo-realistic smart 3D models.

Currently, most AI models are trained on real-world data which is scarce, time consuming to annotate and involves complex challenges around bias and data privacy. Chameleon’s ‘no-code’, ‘self-serve’ platform allows AI developers to create high-quality, precisely annotated and privacy-compliant synthetic training data, more than 50 times faster than sourcing and annotating ‘real-world’ data.

The application pack is currently being used by retail customers to create training data for AI vision systems addressing use-cases including:

• Identify and understand complex human behavior, such as a lost child or a busy loading bay, to ensure customer and employee safety
• Track customer engagement with individual products in dynamic new ways
• Monitor stock levels on shelf and movement of goods from loading bay to instore
• Spill detection and object recognition for enhanced customer safety and more efficient store management
• Monitor and predict traffic flow to maximise parking efficiency, reduce vehicle emissions and enhance customer experience

Watch a video of the Smart Retail application pack in action here: https://vimeo.com/showcase/mindtechretail

Chris Longstaff, VP Product Management at Mindtech said, “The retail industry is realizing the potential of AI to create more intuitive, efficient and safer retail experiences for customers and employees. However, realizing that vision will require smarter AI systems, trained on large volumes of high-quality privacy-compliant data. Chameleon’s new Retail application pack allows retailers to solve their data scarcity with unlimited self-serve synthetic data. Synthetic data produces better results than relying on real-world data alone and it’s delivering a next-generation of smart retail systems that better understand the way humans interact with each other and the world around them.”

Mindtech’s Chameleon platform is already in use by global retail customers and is available for immediate licensing.

Mindtech will be exhibiting at the National Retail Federation (NRF 2022) at booth # 1064 from Jan 16 – 18.

Mindtech Global

Mindtech Global is the developer of the world’s leading end-to-end ‘synthetic’ data creation platform for the training of AI vision systems. The company’s Chameleon platform is a step change in the way AI vision systems are trained, helping computers understand and predict human interactions in applications ranging across retail, smart home, healthcare and smart city.

Mindtech is headquartered in the UK, with operations across the US and Far East and is funded by investors including Mercia, Deeptech Labs and In-Q-Tel.

www.mindtech.global

The post Train Smart Retail AIs 50x Faster With Mindtech’s New Synthetic Data Application Pack appeared first on Edge AI and Vision Alliance.

Tensilica Vision Q7 DSP enables Light’s Clarity Depth Perception Platform to power next-generation ADAS systems with 10X greater performance than quad-core CPU

REDWOOD CITY and SAN JOSE, Calif., January 13, 2022 – Light and Cadence Design Systems, Inc. (Nasdaq: CDNS) today announced that Light has deployed the Cadence^® Tensilica^® Vision Q7 DSP for use in Light’s Clarity Depth Perception Platform, which provides long-range, high-resolution depth perception using industry-standard cameras benefiting advanced driver-assistance systems (ADAS). The Tensilica Vision Q7 DSP provides Light with real-time data processing, ensuring low-latency, high-bandwidth transmission of high-resolution output.

Light’s solution enables machines to see better than humans by using two or more cameras, novel calibration and unique signal processing to provide unprecedented depth quality across the camera’s field of view. Integrating the Tensilica Vision Q7 DSP into Light’s solution enables up to 10X faster processing of measured depth compared to a quad-core CPU, improving real-time accuracy and reliability. The highly configurable and extensible Vision Q7 DSP is accompanied by an array of optimized computer vision libraries, optional accelerators and toolchains that efficiently handle such workloads. Light implemented its proprietary multi-view depth perception algorithms via custom instructions using the Tensilica Instruction Extension (TIE) language, resulting in an additional 4X performance improvement with 3X area/power savings.

For ADAS, meeting functional safety requirements with full ISO 26262 compliance with ASIL-D is crucial. The Tensilica Vision Q7 DSP’s availability as certified IP along with certified toolchains, libraries and a streamlined failure modes, effects and diagnostic analysis (FMEDA) flow were key factors in Light’s selection of the Vision Q7 DSP.

“Light’s solution enables machines to see better than humans, and a key part of that is processing that information quickly enough to allow the machine to react to the world around it,” said Dave Grannan, CEO, Light. “Integrating the Cadence Tensilica Vision Q7 DSP provides us with enhanced signal processing capabilities at low power with low latency, which complement the technology in the platform, helping us to provide accurate and reliable measured depth for future ADAS systems. Additionally, integrating ISO 26262-certified IP such as the Tensilica Vision Q7 DSP is critical when designing automotive-grade solutions.”

“The Tensilica Vision Q7 DSP continues to be the de-facto standard for vision processors in the majority of today’s markets, providing both high performance and a low-power envelope,” said David Glasco, vice president of research and development for Tensilica IP at Cadence. “Light’s innovative approach to depth perception is quite complex and achieves real-time processing with high accuracy by integrating the Vision Q7 DSP into Light’s Clarity Depth Perception Platform. This showcases the Vision Q7 DSP’s versatility and strengthens our footprint in the intelligent automotive space.”

The Tensilica Vision Q7 DSP supports Cadence’s Intelligent System Design strategy by enabling SoC design excellence. For more information on the Vision Q7 DSP, please visit www.cadence.com/go/VisionQ7adas.

About Cadence

Cadence is a pivotal leader in electronic design, building upon more than 30 years of computational software expertise. The company applies its underlying Intelligent System Design strategy to deliver software, hardware and IP that turn design concepts into reality. Cadence customers are the world’s most innovative companies, delivering extraordinary electronic products from chips to boards to systems for the most dynamic market applications, including consumer, hyperscale computing, 5G communications, automotive, mobile, aerospace, industrial and healthcare. For seven years in a row, Fortune magazine has named Cadence one of the 100 Best Companies to Work For. Learn more at cadence.com.

About Light

Light is pioneering real-time 3D perception to redefine how vehicles see the world. By combining the company’s breakthroughs in computational imaging with multi-camera calibration and advanced signal processing, Light’s technology provides accurate depth at both near and far distances in real time. Founded in 2013, Light is headquartered in Redwood City, CA. See more at www.light.co.

The post Light Leverages Cadence Tensilica Vision Q7 DSP for Enhanced Depth Perception in Next-Generation ADAS Systems appeared first on Edge AI and Vision Alliance.

On February 24, 2022 at 9 am PT (noon ET), Rob Telson, Vice President of Worldwide Sales, and Todd Vierra, Director of Customer Engagement, both of BrainChip, will present the free hour webinar “Developing Optimized Systems with BrainChip’s Akida Neuromorphic Processor,” organized by the Edge AI and Vision Alliance. Here’s the description, from the event registration page:

BrainChip’s Akida processor today finds use in a diversity of applications, such as classifying images, identifying odors and tastes, recognizing breath data for disease classification, identifying air quality, interpreting LiDAR laser light data, recognizing keywords, and detecting cybersecurity attacks. Akida leverages advanced neuromorphic computing as its processing “engine”, delivering key features such as one-shot learning and on-device computing with no “cloud” dependencies. As such, it’s particularly valuable in evolving smart “edge” devices, where privacy, security, low power consumption, low latency and high performance, low cost, and small size are all important criteria.

In this session, you’ll learn how to easily develop efficient AI in edge devices by implementing Akida IP either into your SoC or as standalone silicon. The presenters will provide detailed performance and other results, derived from real-life system implementations using production Akida silicon, as well as share a variety of design techniques and support resources. A question-and-answer session will follow the presentation.

To register for this free webinar, please see the event page. For more information, please email webinars@edge-ai-vision.com.

The post Free Webinar Explores How to Develop Optimized Systems Using Neuromorphic Processing appeared first on Edge AI and Vision Alliance.

January 17, 2022— Nota (CEO Myungsu Chae), an AI optimization technology company, announced that it has been promoted to Premier status in the NVIDIA Inception program.

NVIDIA Inception nurtures cutting-edge startups revolutionizing industries with advancements in AI and data science. The free program has 9,000+ members who are given access to the best technical tools, latest resources, and opportunities to connect with investors. As a startup matures, its Inception benefits also evolve to further company growth. Premier members receive increased NVIDIA marketing support, access to Premier-only member events, and a dedicated NVIDIA relationship manager.

Selected as an Inception Premier member through the program’s rigorous selection process, Nota, based on its AI optimization source technology, provides NetsPresso, a representative solution, an edge-based intelligent transportation system, facial recognition-based access authentication, and low-power driver monitoring solution in the vehicle. Nota plans to advance its AI optimization technology and meet customer needs in the market through close collaboration with NVIDIA.

Nota CEO Myungsu Chae said, “I am proud to have been selected alongside outstanding global AI companies in the first year of cooperation with NVIDIA.”

In addition, Nota participated in NVIDIA GTC, held in November, and introduced NetsPresso, a hardware-aware AutoML platform. Nota’s AI optimization technology, edge-based intelligent transportation system, and low-power driver monitoring solution will also be presented at 2022 Embedded Vision Summit, the premier conference and expo devoted to practical, deployable computer vision and visual AI to be held in California, USA in June.

Nota Website : https://nota.ai

The post AI Optimization Technology Company Nota Selected as NVIDIA Inception Premier Member appeared first on Edge AI and Vision Alliance.