Object Detection On Crowdhuman Full Body

Metrics

AP

mMR

Results

Performance results of various models on this benchmark

			Paper Title
InternImage-H	97.2	-	InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutions
MMPedestron	97.1	30.8	When Pedestrian Detection Meets Multi-Modal Learning: Generalist Model and Benchmark Dataset
Progressive DETR	94.1	37.7	Progressive End-to-End Object Detection in Crowded Scenes
DDQ DETR (R50)	93.8	39.7	Dense Distinct Query for End-to-End Object Detection
DDQ R-CNN (R50)	93.5	40.4	Dense Distinct Query for End-to-End Object Detection
Hulk(Finetune, ViT-L)	93	36.5	Hulk: A Universal Knowledge Translator for Human-Centric Tasks
DDQ FCN (R50 One-Stage)	92.7	41.0	Dense Distinct Query for End-to-End Object Detection
UniHCP (finetune)	92.5	41.6	UniHCP: A Unified Model for Human-Centric Perceptions
Hulk(Finetune, ViT-B)	92.4	40.7	Hulk: A Universal Knowledge Translator for Human-Centric Tasks
V2F-Net	91.03	42.28	V2F-Net: Explicit Decomposition of Occluded Pedestrian Detection
CrowdDet	90.7	41.4	Detection in Crowded Scenes: One Proposal, Multiple Predictions
Beta R-CNN	89.6	40.3	Beta R-CNN: Looking into Pedestrian Detection from Another Perspective
NOH-NMS	89.0	43.9	NOH-NMS: Improving Pedestrian Detection by Nearby Objects Hallucination
IterDet (Faster RCNN, ResNet50, 2 iterations)	88.08	49.44	IterDet: Iterative Scheme for Object Detection in Crowded Environments
PS-RCNN (Faster RCNN, ResNet50, COCO Instance Masks	87.94	-	PS-RCNN: Detecting Secondary Human Instances in a Crowd via Primary Object Suppression
PS-RCNN (Faster RCNN, ResNet50)	86.05	-	PS-RCNN: Detecting Secondary Human Instances in a Crowd via Primary Object Suppression
Faster RCNN (ResNet50)	84.95	50.49	CrowdHuman: A Benchmark for Detecting Human in a Crowd
Adaptive NMS (Faster RCNN, ResNet50)	84.71	49.73	Adaptive NMS: Refining Pedestrian Detection in a Crowd
IterDet (Faster RCNN, ResNet50, 1 iteration)	84.43	49.12	IterDet: Iterative Scheme for Object Detection in Crowded Environments

0 of 19 row(s) selected.

Object Detection On Crowdhuman Full Body

Metrics

AP

mMR

Results

Performance results of various models on this benchmark

			Paper Title
InternImage-H	97.2	-	InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutions
MMPedestron	97.1	30.8	When Pedestrian Detection Meets Multi-Modal Learning: Generalist Model and Benchmark Dataset
Progressive DETR	94.1	37.7	Progressive End-to-End Object Detection in Crowded Scenes
DDQ DETR (R50)	93.8	39.7	Dense Distinct Query for End-to-End Object Detection
DDQ R-CNN (R50)	93.5	40.4	Dense Distinct Query for End-to-End Object Detection
Hulk(Finetune, ViT-L)	93	36.5	Hulk: A Universal Knowledge Translator for Human-Centric Tasks
DDQ FCN (R50 One-Stage)	92.7	41.0	Dense Distinct Query for End-to-End Object Detection
UniHCP (finetune)	92.5	41.6	UniHCP: A Unified Model for Human-Centric Perceptions
Hulk(Finetune, ViT-B)	92.4	40.7	Hulk: A Universal Knowledge Translator for Human-Centric Tasks
V2F-Net	91.03	42.28	V2F-Net: Explicit Decomposition of Occluded Pedestrian Detection
CrowdDet	90.7	41.4	Detection in Crowded Scenes: One Proposal, Multiple Predictions
Beta R-CNN	89.6	40.3	Beta R-CNN: Looking into Pedestrian Detection from Another Perspective
NOH-NMS	89.0	43.9	NOH-NMS: Improving Pedestrian Detection by Nearby Objects Hallucination
IterDet (Faster RCNN, ResNet50, 2 iterations)	88.08	49.44	IterDet: Iterative Scheme for Object Detection in Crowded Environments
PS-RCNN (Faster RCNN, ResNet50, COCO Instance Masks	87.94	-	PS-RCNN: Detecting Secondary Human Instances in a Crowd via Primary Object Suppression
PS-RCNN (Faster RCNN, ResNet50)	86.05	-	PS-RCNN: Detecting Secondary Human Instances in a Crowd via Primary Object Suppression
Faster RCNN (ResNet50)	84.95	50.49	CrowdHuman: A Benchmark for Detecting Human in a Crowd
Adaptive NMS (Faster RCNN, ResNet50)	84.71	49.73	Adaptive NMS: Refining Pedestrian Detection in a Crowd
IterDet (Faster RCNN, ResNet50, 1 iteration)	84.43	49.12	IterDet: Iterative Scheme for Object Detection in Crowded Environments

0 of 19 row(s) selected.