Battery-free AI of Things

Extremely low Resolution Camera Tag

action recognition network for tag

Despite the potential of vision-based monitoring, data leakage concerns hinder its wide deployment in personal spaces. Besides, continuous and pervasive monitoring without blind spots in complicated indoor spaces requires a scalable system. Our system presents a vision-based end-to-end action recognition framework that (i) intrinsically achieves data anonymity from the sensing stage and (ii) battery-free operation for blind spot-free continuous monitoring. Our custom tag provides > 2 fps videos even under dark places (e.g., deep in the shelf without direct light), and our system of three tags shows an accuracy of 98.16% on classifying ten actions. It leverages an extremely low-resolution (eLR) near-infrared (NIR) image sensor with 6×10 pixels for video anonymity and an RFID-compliant fully-passive tag for real-time eLR video streaming.

Publications

Seungwoo Shim*, Hyeonho Shin*, Myeongkyun Cho, Youngki Lee, Jinwoo Shin, Song Min Kim (*: equal contribution),  The 22nd ACM/IEEE International Conference on Information Processing in Sensor Networks, (Acceptance Rate 22/83 = 26.5%)[paper] [video(short)]

Members

Hyeonho Shin
Ph.D. Program
hh.shin[at]kaist.ac.kr

Seungwoo Shim
Ph.D. Program
shimsw0608[at]kaist.ac.kr

Namjo Ahn
PhD + MS Joint Program
njahn0716[at]kaist.ac.kr