I2R Researchers Won the IEEE CVPR’21 UG2+ Challenge

I2R Researchers Won 1st Place at IEEE CVPR’21 UG2+ Challenge

Yang Zaifeng (IHPC), together with I2R colleagues (Jin Ruibin, Wu Keyu, Hou Yubo, Chen Zhenghua and Wu Min) formed a team and participated in IEEE CVPR’21 UG2+ Challenge under "A*StarTrek” and emerged winner for Track 2.1: Fully Supervised Action Recognition in the Dark.
This challenge, sponsored by NVIDIA and Walmart, aims to explore possible video-based technologies, without any additional sensors, that are robust to darkness, extracting effective action features from dark videos, which would benefit in various downstream applications. 69 teams enrolled in Track2.1, and 13 teams completed the testing phase.

The team was able to adapt the multi-model based pseudo grouth truth (PGT) generation and achieved an accuracy of 93.7161% in this challenge and took home 1st place.

More details can be found here.