Action Recognition on CMU 2235 sequences (4-fold cross val)
88.3AccuracyClips + CNN + MTLN
Evaluation results
| Method | Links | |
|---|---|---|
| Clips + CNN + MTLNinput=Clips, architecture=MTLN2017.03 | 88.3 | |
| Clips + CNN + Concatenationinput=Clips, aggregation=Concatenation2017.03 | 85.76 | |
| Clips + CNN + Poolinginput=Clips, aggregation=Pooling2017.03 | 85.56 | |
| Frames + CNNinput=Individual Frames2017.03 | 85.36 | |
| Deep LSTM + Co-occurrence2017.03 | 81.04 | |
| Deep LSTM2017.03 | 79.53 | |
| Hierarchical RNN2017.03 | 75.02 | |
| Coordinates + FTPinput=Coordinates, feature=FTP2017.03 | 73.61 |