Action Recognition on UCF11
98.9AccuracyMahmoud & Abdelhakim
Evaluation results
| Method | Links | |
|---|---|---|
| Mahmoud & AbdelhakimApproach=BiLSTM + Attention2026.04 | 98.9 | |
| DualStreamHybrid – AttentionApproach=ViT-Tiny + MobileNetV2, cross-attn2026.04 | 98.12 | |
| DualStreamHybrid – WeightedApproach=ViT-Tiny + MobileNetV2, weighted2026.04 | 96.88 | |
| RGB-only ViT-TinyApproach=ViT-Tiny/16, single stream2026.04 | 95.94 | |
| Gammulle et al.Approach=Two-Stream LSTM (VGG-16)2026.04 | 94.6 | |
| 3DCNNModel compression status=Original, Evaluation protocol=k-fold2019.12 | 93.78 | |
| Fisher Vector (LOG)Evaluation protocol=LOG2019.12 | 93.77 | |
| TT-3DCNNModel compression status=TT-decomposed, Evaluation protocol=k-fold2019.12 | 93.56 | |
| Local Motion (LOG)Evaluation protocol=LOG2019.12 | 88 | |
| TR-LSTM (k-fold)Evaluation protocol=k-fold2019.12 | 86.9 | |
| BT-LSTM (k-fold)Evaluation protocol=k-fold2019.12 | 85.3 | |
| Yang et al.Approach=Block-Term LSTM2026.04 | 85.3 | |
| Visual Attention (k-fold)Evaluation protocol=k-fold2019.12 | 85 | |
| TT-3DCNNModel compression status=TT-decomposed, Evaluation protocol=LOG2019.12 | 83.81 | |
| 3DCNNModel compression status=Original, Evaluation protocol=LOG2019.12 | 83.72 | |
| TT-GRU (k-fold)Evaluation protocol=k-fold2019.12 | 81.3 | |
| Liu et al.Approach=HOG/HOF + SVM2026.04 | 71.2 |