Object Detection on CrowdHuman (full body)

Metric: mMR (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	mMR▼	Extra Data	Paper	Date↕	Code
1	Faster RCNN (ResNet50)	50.49	No	CrowdHuman: A Benchmark for Detecting Human in a...	2018-04-30	Code
2	Adaptive NMS (Faster RCNN, ResNet50)	49.73	No	Adaptive NMS: Refining Pedestrian Detection in a...	2019-04-07	-
3	IterDet (Faster RCNN, ResNet50, 2 iterations)	49.44	No	IterDet: Iterative Scheme for Object Detection i...	2020-05-12	-
4	IterDet (Faster RCNN, ResNet50, 1 iteration)	49.12	No	IterDet: Iterative Scheme for Object Detection i...	2020-05-12	-
5	NOH-NMS	43.9	No	NOH-NMS: Improving Pedestrian Detection by Nearb...	2020-07-27	Code
6	V2F-Net	42.28	No	V2F-Net: Explicit Decomposition of Occluded Pede...	2021-04-07	-
7	UniHCP (finetune)	41.6	No	UniHCP: A Unified Model for Human-Centric Percep...	2023-03-06	Code
8	CrowdDet	41.4	No	Detection in Crowded Scenes: One Proposal, Multi...	2020-03-20	Code
9	DDQ FCN (R50 One-Stage)	41	No	Dense Distinct Query for End-to-End Object Detec...	2023-03-22	Code
10	Hulk(Finetune, ViT-B)	40.7	No	Hulk: A Universal Knowledge Translator for Human...	2023-12-04	Code
11	DDQ R-CNN (R50)	40.4	No	Dense Distinct Query for End-to-End Object Detec...	2023-03-22	Code
12	Beta R-CNN	40.3	No	Beta R-CNN: Looking into Pedestrian Detection fr...	2022-10-23	-
13	DDQ DETR (R50)	39.7	No	Dense Distinct Query for End-to-End Object Detec...	2023-03-22	Code
14	Progressive DETR	37.7	No	Progressive End-to-End Object Detection in Crowd...	2022-03-15	Code
15	Hulk(Finetune, ViT-L)	36.5	No	Hulk: A Universal Knowledge Translator for Human...	2023-12-04	Code
16	MMPedestron	30.8	No	When Pedestrian Detection Meets Multi-Modal Lear...	2024-07-14	Code

#1Faster RCNN (ResNet50)SOTA
50.49
mMR· 2018-04-30
CrowdHuman: A Benchmark for Detecting Human in a Crowd Code
#2Adaptive NMS (Faster RCNN, ResNet50)
49.73
mMR· 2019-04-07
Adaptive NMS: Refining Pedestrian Detection in a Crowd
#3IterDet (Faster RCNN, ResNet50, 2 iterations)
49.44
mMR· 2020-05-12
IterDet: Iterative Scheme for Object Detection in Crowded Environments
#4IterDet (Faster RCNN, ResNet50, 1 iteration)
49.12
mMR· 2020-05-12
IterDet: Iterative Scheme for Object Detection in Crowded Environments
#5NOH-NMS
43.9
mMR· 2020-07-27
NOH-NMS: Improving Pedestrian Detection by Nearby Objects Hallucination Code
#6V2F-Net
42.28
mMR· 2021-04-07
V2F-Net: Explicit Decomposition of Occluded Pedestrian Detection
#7UniHCP (finetune)
41.6
mMR· 2023-03-06
UniHCP: A Unified Model for Human-Centric Perceptions Code
#8CrowdDet
41.4
mMR· 2020-03-20
Detection in Crowded Scenes: One Proposal, Multiple Predictions Code
#9DDQ FCN (R50 One-Stage)
41
mMR· 2023-03-22
Dense Distinct Query for End-to-End Object Detection Code
#10Hulk(Finetune, ViT-B)
40.7
mMR· 2023-12-04
Hulk: A Universal Knowledge Translator for Human-Centric Tasks Code
#11DDQ R-CNN (R50)
40.4
mMR· 2023-03-22
Dense Distinct Query for End-to-End Object Detection Code
#12Beta R-CNN
40.3
mMR· 2022-10-23
Beta R-CNN: Looking into Pedestrian Detection from Another Perspective
#13DDQ DETR (R50)
39.7
mMR· 2023-03-22
Dense Distinct Query for End-to-End Object Detection Code
#14Progressive DETR
37.7
mMR· 2022-03-15
Progressive End-to-End Object Detection in Crowded Scenes Code
#15Hulk(Finetune, ViT-L)
36.5
mMR· 2023-12-04
Hulk: A Universal Knowledge Translator for Human-Centric Tasks Code
#16MMPedestron
30.8
mMR· 2024-07-14
When Pedestrian Detection Meets Multi-Modal Learning: Generalist Model and Benchmark Dataset Code