Video Frame Classification on AVA temporal annotation-only
14.9mAPI3D + 3 TGMs + super-events
Evaluation results
| Method | Links | |
|---|---|---|
| I3D + 3 TGMs + super-eventsSetting=temporal annotation-only, number of TGM layers=3, temporal module=super-events2018.03 | 14.9 | |
| I3D + 3 TGMsSetting=temporal annotation-only, number of TGM layers=32018.03 | 14.5 | |
| I3D + 1 TGMsSetting=temporal annotation-only, number of TGM layers=12018.03 | 11.2 | |
| I3D + super-eventsSetting=temporal annotation-only, temporal module=super-events2018.03 | 9.8 | |
| I3D + 3 temporal conv. layersSetting=temporal annotation-only, number of temporal conv layers=32018.03 | 7.9 | |
| I3D + LSTMSetting=temporal annotation-only, temporal module=LSTM2018.03 | 7.8 | |
| I3D baselineSetting=temporal annotation-only2018.03 | 7.5 | |
| RandomSetting=temporal annotation-only2018.03 | 2.65 |