search

TengdaHan / CoCLR

[NeurIPS'20] Self-supervised Co-Training for Video Representation Learning. Tengda Han, Weidi Xie, Andrew Zisserman.
Apache License 2.0
286 stars 32 forks source link

2-Stream evaluation on UCF101 #34

Closed thematrixduo closed 3 years ago

thematrixduo commented 3 years ago

In your paper (Table 1 and 2) you used RGB+Flow for evaluation on UCF101. I couldn't find this 2-stream evaluation codes anywhere in this repo. I am missing anything? Would be grateful if you can point me to the relevant locations.

TengdaHan commented 3 years ago

Hi. I just uploaded this for reference: https://github.com/TengdaHan/CoCLR/blob/main/eval/merge_2stream_prob.py For each video, what I did is average the probability vectors (e.g. shape=1x101, sum up to 1.0) given by RGB and optical flow stream. then take the argmax as the two-stream prediction.