Crowds on ShanghaiTech B

Metric: MAE (lower is better)

LeaderboardDataset

Loading chart...

Results

Hide extra data

Sort:

#	Model↕	MAE▲	Extra Data	Paper	Date↕	Code
1	EBC-ZIP-B	5.51	No	EBC-ZIP: Improving Blockwise Crowd Counting with...	2025-06-24	Code
2	PSL-Net	5.8	Yes	-	-	-
3	CLIP-EBC (ViT-L/14)	5.9	Yes	CLIP-EBC: CLIP Can Count Accurately through Enha...	2024-03-14	Code
4	CLIP-EBC (ResNet50)	6	Yes	CLIP-EBC: CLIP Can Count Accurately through Enha...	2024-03-14	Code
5	GauNet (ResNet-50)	6	Yes	Rethinking Spatial Invariance of Convolutional N...	2022-06-10	Code
6	SGANet	6.3	Yes	Crowd Counting via Segmentation Guided Attention...	2019-11-18	Code
7	M-SFANet+M-SegNet	6.32	Yes	Encoder-Decoder Based Convolutional Neural Netwo...	2020-03-12	Code
8	SPANet	6.5	Yes	Learning Spatial Awareness to Improve Crowd Coun...	2019-09-16	-
9	CLIP-EBC (ViT-B/16)	6.6	Yes	CLIP-EBC: CLIP Can Count Accurately through Enha...	2024-03-14	Code
10	SGANet + CL	6.6	Yes	Crowd Counting via Segmentation Guided Attention...	2019-11-18	Code
11	S-DCNet	6.7	Yes	From Open Set to Closed Set: Counting Objects by...	2019-08-15	Code
12	FusionCount	6.9	No	FusionCount: Efficient Crowd Counting via Multis...	2022-02-28	Code
13	CSRNet-EBC	6.9	No	CLIP-EBC: CLIP Can Count Accurately through Enha...	2024-03-14	Code
14	DMCount-EBC	7	No	CLIP-EBC: CLIP Can Count Accurately through Enha...	2024-03-14	Code
15	DM-Count	7.4	No	Distribution Matching for Crowd Counting	2020-09-28	Code
16	CAN	7.8	No	Context-Aware Crowd Counting	2018-11-26	Code
17	LSC-CNN	8.1	No	Locate, Size and Count: Accurately Resolving Peo...	2019-06-18	Code
18	SANet	8.4	No	-	-	Code
19	APGCC	8.7	No	Improving Point-based Crowd Counting and Localiz...	2024-05-17	Code
20	OrdinalEntropy	9.1	No	Improving Deep Regression with Ordinal Entropy	2023-01-21	Code
21	SAFECount	9.98	No	Few-shot Object Counting with Similarity-Aware F...	2022-01-22	Code
22	CSRNet	10.6	No	CSRNet: Dilated Convolutional Neural Networks fo...	2018-02-27	Code
23	ic-CNN	10.7	No	Iterative Crowd Counting	2018-07-26	-
24	IG-CNN	13.6	No	Divide and Grow: Capturing Huge Diversity in Cro...	2018-07-26	-
25	Liu et al.	13.7	No	Leveraging Unlabeled Data for Crowd Counting by ...	2018-03-08	Code
26	ACSCP	17.2	No	-	-	Code
27	D-ConvNet	18.7	No	-	-	Code
28	Cascaded-MTL	20	No	CNN-based Cascaded Multi-task Learning of High-l...	2017-07-30	Code
29	CP-CNN	20.1	No	Generating High-Quality Crowd Density Maps using...	2017-08-02	-
30	Switch-CNN	21.6	No	Switching Convolutional Neural Network for Crowd...	2017-08-01	Code
31	MCNN	26.4	No	-	-	Code
32	Zhang et al.	32	No	-	-	-

#1EBC-ZIP-BSOTA
5.51
MAE· 2025-06-24
EBC-ZIP: Improving Blockwise Crowd Counting with Zero-Inflated Poisson Regression Code
#2PSL-Net
5.8
MAE· Extra Data
No paper
#3CLIP-EBC (ViT-L/14)SOTA
5.9
MAE· Extra Data· 2024-03-14
CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification Code
#4CLIP-EBC (ResNet50)
6
MAE· Extra Data· 2024-03-14
CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification Code
#5GauNet (ResNet-50)SOTA
6
MAE· Extra Data· 2022-06-10
Rethinking Spatial Invariance of Convolutional Networks for Object Counting Code
#6SGANetSOTA
6.3
MAE· Extra Data· 2019-11-18
Crowd Counting via Segmentation Guided Attention Networks and Curriculum Loss Code
#7M-SFANet+M-SegNet
6.32
MAE· Extra Data· 2020-03-12
Encoder-Decoder Based Convolutional Neural Networks with Multi-Scale-Aware Modules for Crowd Counting Code
#8SPANetSOTA
6.5
MAE· Extra Data· 2019-09-16
Learning Spatial Awareness to Improve Crowd Counting
#9CLIP-EBC (ViT-B/16)
6.6
MAE· Extra Data· 2024-03-14
CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification Code
#10SGANet + CL
6.6
MAE· Extra Data· 2019-11-18
Crowd Counting via Segmentation Guided Attention Networks and Curriculum Loss Code
#11S-DCNetSOTA
6.7
MAE· Extra Data· 2019-08-15
From Open Set to Closed Set: Counting Objects by Spatial Divide-and-Conquer Code
#12FusionCount
6.9
MAE· 2022-02-28
FusionCount: Efficient Crowd Counting via Multiscale Feature Fusion Code
#13CSRNet-EBC
6.9
MAE· 2024-03-14
CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification Code
#14DMCount-EBC
7
MAE· 2024-03-14
CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification Code
#15DM-Count
7.4
MAE· 2020-09-28
Distribution Matching for Crowd Counting Code
#16CANSOTA
7.8
MAE· 2018-11-26
Context-Aware Crowd Counting Code
#17LSC-CNN
8.1
MAE· 2019-06-18
Locate, Size and Count: Accurately Resolving People in Dense Crowds via Detection Code
#18SANet
8.4
MAE
No paperCode
#19APGCC
8.7
MAE· 2024-05-17
Improving Point-based Crowd Counting and Localization Based on Auxiliary Point Guidance Code
#20OrdinalEntropy
9.1
MAE· 2023-01-21
Improving Deep Regression with Ordinal Entropy Code
#21SAFECount
9.98
MAE· 2022-01-22
Few-shot Object Counting with Similarity-Aware Feature Enhancement Code
#22CSRNetSOTA
10.6
MAE· 2018-02-27
CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes Code
#23ic-CNN
10.7
MAE· 2018-07-26
Iterative Crowd Counting
#24IG-CNN
13.6
MAE· 2018-07-26
Divide and Grow: Capturing Huge Diversity in Crowd Images with Incrementally Growing CNN
#25Liu et al.
13.7
MAE· 2018-03-08
Leveraging Unlabeled Data for Crowd Counting by Learning to Rank Code
#26ACSCP
17.2
MAE
No paperCode
#27D-ConvNet
18.7
MAE
No paperCode
#28Cascaded-MTLSOTA
20
MAE· 2017-07-30
CNN-based Cascaded Multi-task Learning of High-level Prior and Density Estimation for Crowd Counting Code
#29CP-CNN
20.1
MAE· 2017-08-02
Generating High-Quality Crowd Density Maps using Contextual Pyramid CNNs
#30Switch-CNN
21.6
MAE· 2017-08-01
Switching Convolutional Neural Network for Crowd Counting Code
#31MCNN
26.4
MAE
No paperCode
#32Zhang et al.
32
MAE
No paper