Program

Schedule

The workshop will be on 08:50-18:45 PM CDT (GMT-5) on June 20, 2022.

8:50-9:00 Welcome by organizers: Peter Vajda (Meta), Bichen Wu (Meta), Peizhao Zhang (Meta), Andrew Howard (Google), Grace Chu (Google), Yaeyoun Kim (Google), Pete Warden (Google), Chris Rowen (Cisco), Yung-Hsiang Lu (Purdue Univ.), Kurt Keutzer (Berkeley)

09:00-09:30 Invited talk: "Efficient and Robust Fully-attentional Networks" by Jiashi Feng (Remote)

09:30-10:00 Invited talk: "Towards Compact and Tiny AI models on Edge" by Yiran Chen

10:00-10:15 Oral paper presentation: Hybrid Consistency Training with Prototype Adaptation for Few-Shot Learning

10:15-10:30 Oral paper presentation: YOLO-Pose: Enhancing YOLO for Multi Person Pose Estimation Using Object Keypoint Similarity Loss

10:30-10:45 Oral paper presentation: DA3: Dynamic Additive Attention Adaption for Memory-Efficient On-Device Learning (Remote)

10:45-11:00 Oral paper presentation: Cyclical Pruning for Sparse Neural Networks (Remote)

11:00-11:30 Invited talk: "Researching Efficient networks for hardware that does not exist yet: physics and economics of ML co-design" by Łukasz Lew

11:30-12:00 Invited talk by Richard Newcombe

12:00-13:00 Lunch break

13:00-13:15 Oral paper presentation: Conjugate Adder Net (CAddNet) - a Space-Efficient Approximate CNN

13:15-13:30 Oral paper presentation: Simulated Quantization, Real Power Savings

13:30-13:45 Oral paper presentation: Event Transformer. A sparse-aware solution for efficient event data processing

13:45-14:00 Oral paper presentation: Searching for Efficient Neural Architectures for On-Device ML on Edge TPUs

14:00-14:30 Invited talk: "HyperTransformer: generating tiny models in a single shot" by Mark Sandler

14:30-15:00 Invited talk: "Tackling Model and Data Efficiency Challenges for Computer Vision" by Bichen Wu

15:00-16:00 Poster session

16:00-16:15 "Designing A Low-Power Video Recognition System", by Zhangyang Atlas Wang, Assistant Professor, UT Austin. 1st award of Low-Power Computer Vision (LPCV) challenge, video track

16:15-16:30 "Compressing models with few samples", by Junjie Liu, Senior Research Engineer, Meituan Research. 2nd award of LPCV challenge, video track. (Remote)

16:30-16:45 "When industrial model toolchain meets Xilinx FPGA", by Jiahao Hu, Ruihao Gong (龚睿昊), SenseTime Research. 1st award of LPCV challenge, FPGA track (Remote)

16:45-17:00 "Enable Efficient Deep Learning on Mobile Devices", by Song Han, Assistant Professor, MIT. 2nd award of LPCV challenge, FPGA track (Remote)

17:00-17:45 Panel discussion on new opportunities and challenges of efficient computer vision.

17:45-18:00 Closing remarks and announcing the Best Paper award.