Optimizing SSD Object Detection for Low-power Devices

Tuesday, May 21, 1:35 PM - 2:05 PM
Summit Track: 
Technical Insights I
Mission City B1-B5

Deep learning-based computer vision models have gained traction in applications requiring object detection, thanks to their accuracy and flexibility. For deployment on low-power hardware, single-shot detection (SSD) models are attractive due to their speed when operating on inputs with small spatial dimensions.

The key challenge in creating efficient embedded implementations of SSD is not in the feature extraction module, but rather is due to the non-linear bottleneck in the detection stage, which does not lend itself to parallelization. This hinders the ability to lower the processing time per frame, even with custom hardware. We will describe in detail a data-centric optimization approach to SSD. Our approach drastically lowers the number of priors (“anchors”) needed for the detection, and thus linearly decreases time spent on this costly part of the computation. Thus, specialized processors and custom hardware may be better utilized, yielding higher performance and lower latency regardless of the specific hardware used.


Moses Guttmann

CTO and Founder, allegro.ai

Moses Guttmann is CTO and founder of allegro.ai, a company building a holistic deep learning platform. A 15-year computer vision and deep learning specialist, Moses is a seasoned business and product leader with a two-decade track record in leading and driving execution of large-scale, complex products in a multitude of disciplines. Previously, he founded an innovative semiautomatic 3D conversion company, built face recognition technologies and implemented wavelet-based compression on embedded systems for several companies. He holds an MSc in computer science (cum laude) from Tel Aviv University.

See you at the Summit! May 20-23 in Santa Clara, California!
Register today and reserve your hotel room!