Temporal Saliency Query Network for Efficient Video Recognition
The first work to model temporal sampling as a query-response task
Accepted by ECCV2022 (19.9$\%$ acceptance rate).
Publications
NSNet: Non-saliency Suppression Sampler for Efficient Video Recognition
A sampler with a $4\times$ faster practical speed than SOTA methods.
Accepted by ECCV2022 (19.9$\%$ acceptance rate).