Action Recognition on KTH (Full Dataset)
99.3AccuracyMulti-class SVM + HOG + LBP + texture features
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Multi-class SVM + HOG + LBP + texture features2026.05 | 99.3 | 10 | |
| 26-layer CNN + PDaUM preprocessingNetwork Size=~195M2026.05 | 98.3 | — | |
| GRNN + motion tracking + GMM/KF2026.05 | 96.3 | — | |
| Bio-inspired SNN + preprocessingNetwork Size=24,0002026.05 | 92.3 | — | |
| TDRC + silhouette segmentation, keyframe selection, HOG, PCANetwork Size=6002026.05 | 90.83 | 160 |