Crowd Event Perception Based on Spatiotemporal Weber Field

Table 1

Comparison of precision and recall for different algorithms.

Index	Run (PETS209)		Run (UMN)		Gather		Split		Evacuation		Opposite flow
	Precision	Recall	Precision	Recall	Precision	Recall	Precision	Recall	Precision	Recall	Precision	Recall
Proposed method	0.887	0.902	0.847	0.902	0.825	0.883	0.912	0.882	0.903	0.878	0.879	0.854
Social force [24]	0.812	0.887	0.792	0.891	0.767	0.895	0.807	0.879	0.781	0.882	0.817	0.832
HOG/HOF [36]	0.892	0.919	0.832	0.878	0.817	0.836	0.923	0.813	0.915	0.867	0.833	0.781
HOG3D [37]	0.803	0.786	0.812	0.793	0.803	0.784	0.827	0.832	0.831	0.852	0.821	0.771
Optical flow [8]	0.836	0.868	0.803	0.822	0.809	0.812	0.903	0.792	0.822	0.841	0.812	0.769