Visual Interaction Force Estimation Based on Time-Sensitive Dual-Resolution Learning Network

Table 2

Comparison results with state-of-the-art spatiotemporal methods.

Method	RMSE	MAE	MSE	R2	Inference time (s)
C3D-Desnet3D (, 400000 images)	0.1541	0.0929	0.0237	0.5797
T3D-EfficientnetB0 (, 400000 images)	0.1652	0.0768	0.0273	0.4491
T3D-Resnet50 (, 400000 images)	0.1225	0.0652	0.0150	0.7926
Our method (, 400000 images)	0.0397	0.0243	0.0015	0.9725
Simplified version of our method (, 400000 images)	0.0313	0.0183	0.0009	0.9833