CrowdPose: Efficient Crowded Scenes Pose Estimation and a New Benchmark

By Jiefeng Li, Can Wang, Bernie Hao Zhu, Yihuan Mao, Haoshu Fang, Cewu Lu. Computer Vision and Pattern Recognition, 2018

627 citations recorded by the source corpus.

Multi-person pose estimation is fundamental to many computer vision tasks and has made significant progress in recent years. However, few previous methods explored the problem of pose estimation in crowded scenes while it remains challenging and inevitable in many scenarios. Moreover, current benchmarks cannot provide an appropriate evaluation for such cases. In this paper, we propose a novel and efficient method to tackle the problem of pose estimation in the crowd and a new dataset to better evaluate algorithms. Our model consists of two key components: joint-candidate single person pose estimation (SPPE) and global maximum joints association. With multi-peak prediction for each joint and global association using the graph model, our method is robust to inevitable interference in crowded scenes and very efficient in inference. The proposed method surpasses the state-of-the-art methods on CrowdPose dataset by 5.2 mAP and results on MSCOCO dataset demonstrate the generalization ability of our method.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/15c7357994e92b5f617da1c6f1220f5251166890

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Deep Learning-based Human Pose Estimation: A Survey (2020, methodology): “We refer the readers to the original references for details about other datasets including LSP [90], FLIC [213], AIC-HKD [257], CrowdPose [114], Penn Action [291], J-HMDB [83], and HiEve [133].”
FineGym: A Hierarchical Video Dataset for Fine-Grained Action Understanding (2020, background): “Figure 5: Top-row represents the results of person detection and pose estimation [12, 27] for a Vault routine, and the bottomrow visualizes the optical flow [51] features.”
GHUM & GHUML: Generative 3D Human Shape and Articulated Pose Models (2020, methodology): “Several models rely on feed-forward pose and shape prediction based on different learning architectures, on pose prediction followed by pose and shape refinement to body joints of semantic body part segmentations, or on multicamera fusion[28, 43, 44, 20, 36, 5, 18].”
HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation (2019, methodology): “HigherHRNet also outperforms the previous best method [24] (which performs a global reﬁnement of top-down method [15]) by a healthy margin of 1.6 AP and most of the gain comes from AP M (+1.8 AP) and AP H (+1.5 AP), which contains images with the most crowd.”
High-Order Information Matters: Learning Relation and Topology for Occluded Person Re-Identification (2020, background): “Please note that although the human key-points prediction have achieved high accuracy, they still suffer from unsatisfying performance under occluded/partial images [17].”
RTMPose: Real-Time Multi-Person Pose Estimation based on MMPose (2023, methodology): “CrowdPose [32] SimpleBaseline [59] ResNet-50 256× 192 5.”
Multi-animal pose estimation, identification and tracking with DeepLabCut (2022, background): “33 but its computation was adapted to keypoints.”
Human Pose Regression with Residual Log-likelihood Estimation (2021, background): “Many works [45, 14, 12, 64, 36, 27, 52] extend this idea to multi-person pose estimation following the top-down framework, i.”
Monocular, One-stage, Regression of Multiple 3D People (2020, background): “6M [14] and MPI-INF-3DHP [34]), one pseudo-label 3D dataset (UP [26]) and four in-the-wild 2D pose datasets (MS COCO [30], MPII [2], LSP [16, 17] and AICH [50]).”
Whole-Body Human Pose Estimation in the Wild (2020, background): “There have been several body pose datasets [2,3,10,30,31,62].”
InstaBoost: Boosting Instance Segmentation via Probability Map Guided Copy-Pasting (2019, methodology): “[23] proposed a two-step model Mask R-CNN that extends Faster R-CNN framework with a mask head and achieves state-of-the-art on instance segmentation [2] and pose estimation [27] tasks.”
AlphaPose: Whole-Body Regional Multi-Person Pose Estimation and Tracking in Real-Time (2022, background): “MPII [15] 16 ✓ ✓ 40K CrowdPose [19] 14 ✓ ✓ 80K PoseTrack [68] 15 ✓ ✓ 150K COCO [14] 17 ✓ ✓ 250K OneHand10K [70] 21 ✓ ✓ 10K FreiHand [62] 21 ✓ 130K MHP [71] 21 ✓ 80K WFLW [72] 98 ✓ ✓ 10K AFLW [73] 19 ✓ ✓ 25K COFW [74] 29 ✓ ✓ 1852 300W [61] 68 ✓ ✓ 3837 COCO-WholeBody [18] 133 ✓ ✓ ✓ ✓ 250K Halpe-FullBo”