Table 8 Comparison of the confusion between ResNet34-3DRes18 and Res34-SE-IM-Net
From: Fused behavior recognition model based on attention mechanism
Confusing actions | ResNet34-3DRes18 (16frames) | Res34-SE-IM-Net (16 frames) |
|---|---|---|
(flic-flac, cartwheel) | 43% | 30% (−13%) |
(wave, clap) | 32% | 11% (−21%) |
(laugh, smile) | 37% | 16% (−21%) |
(fencing, sword) | 40% | 37% (−3%) |
(cartwheel, handstand) | 26% | 24% (−2%) |