</> Code with Papers by Syntology

Datasets › OCHuman

OCHuman

Introduced by Song-Hai Zhang et al. in Pose2Seg: Detection Free Human Instance Segmentation archive 2025-07-28

This dataset focuses on heavily occluded human with comprehensive annotations including bounding-box, humans pose and instance mask. This dataset contains 13,360 elaborately annotated human instances within 5081 images. With average 0.573 MaxIoU of each person, OCHuman is the most complex and challenging dataset related to human.

Source: https://github.com/liruilong940607/OCHumanApi Image Source: https://github.com/liruilong940607/OCHumanApi

HomepageBenchmarksPapers JSON

Benchmarks archive 2025-07-28

All 6 leaderboards whose dataset resolves to this page shown (sort by any header). "First row" is the archive's own first row at snapshot, in the archive's row order; nothing here re-ranks and metric direction is not asserted.

First row (archive order)PaperCode
Pose Estimation OCHuman ViTPose (ViTAE-G, GT bounding boxes) Test AP 93.3 ViTPose: Simple Vision Transformer Baselines for Human... huggingface/transformers +5 19 Compare
Human Instance Segmentation OCHuman BBox-Mask-Pose 2x AP 32.4 Detection, Pose Estimation and Segmentation for Multiple... MiraPurkrabek/BBoxMaskPose 18 Compare
2D Human Pose Estimation OCHuman BBox-Mask-Pose 2x Test AP 48.3 Detection, Pose Estimation and Segmentation for Multiple... MiraPurkrabek/BBoxMaskPose 11 Compare
Keypoint Detection OCHuman BBox-Mask-Pose 2x Test AP 48.3 Detection, Pose Estimation and Segmentation for Multiple... MiraPurkrabek/BBoxMaskPose 10 Compare
Multi-Person Pose Estimation OCHuman MIPNet (gt-bb) AP50 89.7 Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation rawalkhirodkar/MIPNet 8 Compare
Pose-Based Human Instance Segmentation OCHuman Pose2Seg (plus ground-truth keypoints) AP 55.2 Pose2Seg: Detection Free Human Instance Segmentation open-mmlab/mmpose +6 1 Compare

Papers archive 2025-07-28

29 shown of 29 papers with a leaderboard row on this dataset's benchmarks, newest first. The archive's own "papers using this dataset" list was never published, so this is the benchmark-backed subset; the archive's count for this dataset is 66. The Syntology column is from Syntology's graph (read 2026-09-24), stated per sample; it is not part of any archive number.

DateSamples run Syntology
PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation 1 1 23 May 2025 not harvested
Detection, Pose Estimation and Segmentation for Multiple Bodies: Closing the Virtuous Circle 1 6 2 Dec 2024 not harvested
Crowd-SAM: SAM as a Smart Annotator for Object Detection in Crowded Scenes 1 1 16 Jul 2024 not harvested
You Only Learn One Query: Learning Unified Human Query for Single-Stage Multi-Person Multi-Task Human-Centric Perception 1 3 9 Dec 2023 not harvested
Rethinking pose estimation in crowds: overcoming the detection information-bottleneck and ambiguity 1 1 13 Jun 2023 ran 4 of 6 samples (2 unverified)
RTMPose: Real-Time Multi-Person Pose Estimation based on MMPose 1 1 13 Mar 2023 not harvested
Object-Centric Multi-Task Learning for Human Instances 0 3 13 Mar 2023 not harvested
UniHCP: A Unified Model for Human-Centric Perceptions 1 1 6 Mar 2023 ran 7 of 13 samples (6 unverified)
SEFD: Learning to Distill Complex Pose and Occlusion 1 1 1 Jan 2023 not harvested
Test-time Adaptation vs. Training-time Generalization: A Case Study in Human Instance Segmentation using Keypoints Estimation 0 1 12 Dec 2022 not harvested
Humans need not label more humans: Occlusion Copy & Paste for Occluded Human Instance Segmentation 1 1 7 Oct 2022 ran 0 of 4 samples (4 unverified)
Occlusion-Aware Instance Segmentation via BiLayer Network Architectures 1 1 8 Aug 2022 not harvested
I^2R-Net: Intra- and Inter-Human Relation Network for Multi-Person Pose Estimation 1 1 22 Jun 2022 not harvested
ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation 6 1 26 Apr 2022 ran 18 of 31 samples (13 unverified; 6 pointer-only for licence)
Contextual Instance Decoupling for Robust Multi-Person Pose Estimation 1 1 1 Jan 2022 not harvested
HRFormer: High-Resolution Transformer for Dense Prediction 1 1 18 Oct 2021 ran 7 of 11 samples (4 unverified)
Real-time Human-Centric Segmentation for Complex Video Scenes 1 3 16 Aug 2021 not harvested
Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation 1 7 27 Jan 2021 ran 1 of 1 samples (0 unverified)
TransPose: Keypoint Localization via Transformer 1 2 28 Dec 2020 not harvested
Count- and Similarity-aware R-CNN for Pedestrian Detection 0 2 1 Aug 2020 not harvested
Differentiable Hierarchical Graph Grouping for Multi-Person Pose Estimation 0 3 23 Jul 2020 not harvested
PoSeg: Pose-Aware Refinement Network for Human Instance Segmentation 0 2 7 Jan 2020 not harvested
Single-Stage Multi-Person Pose Machines 1 1 24 Aug 2019 not harvested
CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmark 4 1 2 Dec 2018 not harvested
Simple Baselines for Human Pose Estimation and Tracking 27 7 17 Apr 2018 not harvested
Pose2Seg: Detection Free Human Instance Segmentation 7 4 28 Mar 2018 ran 0 of 16 samples (16 unverified)
Mask R-CNN 179 1 20 Mar 2017 ran 42 of 140 samples (98 unverified; 23 pointer-only for licence)
RMPE: Regional Multi-person Pose Estimation 14 3 1 Dec 2016 not harvested
Associative Embedding: End-to-End Learning for Joint Detection and Grouping 5 6 16 Nov 2016 ran 0 of 8 samples (8 unverified)

Dataset loaders archive 2025-07-28

open-mmlab/mmposeopen-mmlab/mmposepytorch
liruilong940607/OCHumanApi

2 loaders as listed in the archive; links are outbound and not re-checked here.

Tasks archive 2025-07-28

Pose Estimation2D Human Pose EstimationMulti-Person Pose EstimationKeypoint DetectionHuman Instance SegmentationPose-Based Human Instance Segmentation

License archive 2025-07-28

No licence recorded in the archive. Absence here is not a statement about the dataset's terms.

Modalities archive 2025-07-28

Images

Languages archive 2025-07-28

No language tagged.

Variants archive 2025-07-28

  • OCHuman

1 variant name, as the archive lists them.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections