Edge AI and Vision Insights: May 4, 2021 Edition

LETTER FROM THE EDITOR

Dear Colleague,

The Embedded Vision Summit is quickly approaching (May 25-28), and I just took a look at our long (and growing) list of exhibitors. All I can say is wow! These top suppliers are creating cutting-edge building-block technologies that are accelerating the future of edge AI and vision, and they’ll help you find the capabilities you’re looking for so that you can gain a competitive edge.

At the Summit, you can learn more about them and connect by:

Watching demos and seeing the suppliers reveal their latest processors, software, development tools and much more
Scheduling 1:1 meetings with their tech experts to see how their technologies will work for you, and
Visiting their kiosks to gather the information you need.

Whether you’re looking for embedded processors, AI accelerators, software tools, FPGAs, libraries, algorithms, boards, modules, licensable IP, or services, these exhibitors have you covered. Come check out their latest products and meet your next game-changing supplier! See the full list of exhibitors. Then check out the entire event program. And then register for the Summit today!

Brian Dipert
Editor-In-Chief, Edge AI and Vision Alliance

ADDRESSING ENVIRONMENTAL CORNER CASES

Tackling Extreme Visual Conditions for Autonomous UAVs In the Wild
Skydio ships autonomous robots that are flown at scale in complex, unknown environments every day to capture incredible video, automate dangerous inspections and save lives of first responders. They must make decisions at high speed using just their onboard cameras and algorithms running on low-cost hardware. The company has invested five years of R&D into handling extreme visual scenarios not typically considered by academia nor encountered by cars, ground-based robots or AR applications. Drones are commonly used in scenes with few or no semantic priors on the environment and must deftly navigate in the presence of thin objects, extreme lighting, camera artifacts, motion blur, textureless surfaces, vibrations, dirt, smudges and fog. Because photometric signals are not consistent, these challenges are daunting for both classical vision approaches and unsupervised learning. And there is no ground truth for direct supervision. In this presentation, Hayk Martiros, Head of Autonomy at Skydio, takes a detailed look at these issues and how his company tackled them to push autonomous flight into production.

Designing Cameras to Detect the “Invisible”: Handling Edge Cases Without Supervision
Vision systems play an essential role in safety-critical applications, such as advanced driver assistance systems, autonomous vehicles, video security, and fleet management. However, today’s imaging and vision stacks fundamentally rely on supervised training, making it challenging to handle the “edge cases” with datasets that are naturally biased—for example unusual scenes, object types, and environmental conditions, such as rare, dense fog and snow. In this talk, Felix Heide, Co-Founder and Chief Technology Officer at Algolux, introduces the computational imaging and computer vision approaches being used by Algolux to handle such edge cases. Instead of relying purely on supervised downstream networks to become more robust by seeing more training data, Algolux rethinks the camera design itself and optimizes new processing stacks, from photon to detection, that jointly solve this problem. Heide shows how such co-designed cameras, using the company’s Eos camera stack, outperform public and commercial vision systems. He also shows how the same approach can be applied to depth imaging, allowing Algolux to extract accurate, dense depth via low-cost CMOS gated imagers (in the process beating scanning lidar).

VISION FOR MANUFACTURING

Smarter Manufacturing with Deep Learning-Based Machine Vision
As demand for smarter and more efficient manufacturing is growing, IoT technologies⁠—including sensors, edge devices, gateways, servers and the cloud⁠—are being used throughout the factory to compute deep learning analytics workloads at the appropriate location. Efficient data-driven manufacturing can help to reduce labor costs, increase quality and maximize profit. The biggest hindrance to achieving these outcomes is the difficulty in extracting data from vendor-locked and proprietary systems for analytics downstream.

In this presentation, Tara Thimmanaik, Solutions Architect at Intel, covers Intel’s approach to developing open, flexible and scalable solutions, including:

Intel’s technologies such as OpenVINO, Movidius Vision Processor Units, Edge Insights Software (EIS) and deep learning algorithms
How Intel’s offerings come together in the industrial marketplace with partnerships forged to address the constraints of manufacturing infrastructure
Real-world examples highlighting defect detection in textile printing (where 90% accuracy at 50 fps was achieved) and smartphone screen production (where false negatives were only 0.6%)

High-Bandwidth Multicamera Systems with PCIe Backbone
In this presentation, XIMEA’S Kevin Toerne explores high-bandwidth cameras and their application to multi-camera and embedded systems. He presents solutions for machine vision applications, focused on the use of PCIe as a camera interface for embedded and untethered applications. He examines the current and future state of PCIe speeds and connectivity, with examples of available and upcoming hardware. Camera users implementing high-speed or multi-camera applications will learn how a PCIe interface is useful for design, build and implementation of these systems.