How Image Sensor and Video Compression Parameters Impact Vision Algorithms
Recent advances in deep learning algorithms have brought automated object detection and recognition to human accuracy levels on various test datasets. But algorithms that work well on an engineer’s PC often fail when deployed as part of a complete embedded system. In this talk, we’ll examine some of the key embedded vision system elements that can degrade the performance of vision algorithms. For example, in many systems video is compressed, transmitted, and then decompressed before being presented to vision algorithms. Not surprisingly, video encoding parameters, such as bit rate, can have a significant impact on vision algorithm accuracy. Similarly, image sensor parameters can have a profound effect on the nature of the images captured, and therefore on the performance of vision algorithms. We’ll explore how image sensor and video compression parameters impact vision algorithm performance, and discuss methods for selecting the best parameters to aid vision algorithm accuracy.
Ilya Brailovskiy is a Principal Engineer in the Computer Vision and Imaging group at Amazon Lab126. Before joining Amazon, Mr. Brailovskiy worked in architecture, engineering and research leadership roles in Intel’s Graphics Media, Media and Display Platforms, Media Software and Integrated Performance Primitives (IPP) teams. Prior to Intel, Mr. Brailovskiy worked with various companies to design, implement and optimize imaging, signal processing, artificial intelligence and compression algorithms. He holds a Ph.D. in Applied Mathematics, and has authored more than 15 published papers and more than 40 issued or pending patents. Mr. Brailovskiy is Amazon’s representative to the Alliance for Open Media, co-chairing AV1 testing group. When not at work, you may see Ilya on hiking trails in the SF Bay Area and vicinity.