DistInit: Learning Video Representations Without a Single Labeled Video
Open paper
Benchmark
Model
Rank
Results
action-recognition-in-videos-on-hmdb-51
R(2+1)D-18 (DistInit pretraining)
–
Average accuracy of 3 splits: 54.8
action-recognition-in-videos-on-ucf101
R(2+1)D-18 (DistInit pretraining)
–
3-fold Accuracy: 85.8
Rank counts only results with a code link.