Low-Rank and Sparse Based Deep-Fusion Convolutional Neural Network for Crowd Counting

Table 2

Performance comparison of crowd counting methods for Shanghaitech Part A dataset.

Network	Part A dataset
	MAE	MSE	MRE
ACF [3]	390.5	526.8	84.7
LBP + RR	303.2	371.0	70.4
LBP + LSSVM	224.5	294.6	83.3
Patch-CNN [11]	179.7	252.9	67.7
MCNN [12]	110.2	173.2	37.9
Patch-count CNN [25]	118.4	171.1	39.4
Patch-multitask CNN [26]	110.1	170.1	37.0
TSCCM [27]	115.8	167.9	38.5
Long-short CNN [13]	99.1	145.3	30.0
Hydra-CNN [14]	95.7	143.2	26.1
Deep-fusion network	97.9	145.1	29.5
Fusion + RR	126.9	187.8	31.7
Fusion + LSSVM	99.3	145.2	29.1
LFCNN	89.2	141.9	17.3