Location-free Human Pose Estimation
Not on the map. Why?
Abstract (arXiv)
Human pose estimation (HPE) usually requires large-scale training data to reach high performance. However, it is rather time-consuming to collect high-quality and fine-grained annotations for human body. To alleviate this issue, we revisit HPE and propose a location-free framework without supervision of keypoint locations. We reformulate the regression-based HPE from the perspective of classification. Inspired by the CAM-based weakly-supervised object localization, we observe that the coarse keypoint locations can be acquired through the part-aware CAMs but unsatisfactory due to the gap between the fine-grained HPE and the object-level localization. To this end, we propose a customized transformer framework to mine the fine-grained representation of human context, equipped with the structural relation to capture subtle differences among keypoints. Concretely, we design a Multi-scale Spatial-guided Context Encoder to fully capture the global human context while focusing on the part-aware regions and a Relation-encoded Pose Prototype Generation module to encode the structural relations. All these works together for strengthening the weak supervision from image-level category labels on locations. Our model achieves competitive performance on three datasets when only supervised at a category-level and importantly, it can achieve comparable results with fully-supervised methods with only 25% location labels on MS-COCO and MPII.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 49 of its 55 references that are in the map. First posted on arXiv 25 May 2022.
Builds on
49 of 55 references.
References · 49
- 2017Attention Is All You Needcited by 60,361
- 2014Adam: A Method for Stochastic Optimizationcited by 37,240
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
- 2014Microsoft COCO: Common Objects in Contextcited by 17,608
- 2020End-to-End Object Detection with Transformerscited by 7,414
- 2017Mask R-CNNcited by 7,277
- 2020Deformable DETR: Deformable Transformers for End-to-End Object Detectioncited by 3,422
- 2019CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Featurescited by 2,884
- 2015Learning Deep Features for Discriminative Localizationcited by 2,436
- 2020Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformerscited by 1,369
- 2019Deep High-Resolution Representation Learning for Human Pose Estimationcited by 1,339
- 2018OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fieldscited by 1,162
37 more
- 2016Stacked Hourglass Networks for Human Pose Estimationcited by 915
- 2018Simple Baselines for Human Pose Estimation and Trackingcited by 502
- 2017Cascaded Pyramid Network for Multi-Person Pose Estimationcited by 341
- 2016Convolutional Pose Machinescited by 325
- 20213D Human Pose Estimation with Spatial and Temporal Transformerscited by 272
- 2019HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimationcited by 247
- 2018CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmarkcited by 214
- 2016Associative Embedding: End-to-End Learning for Joint Detection and Groupingcited by 212
- 2017Hide-and-Seek: Forcing a Network to be Meticulous for Weakly-supervised Object and Action Localizationcited by 207
- 2019Semantic Graph Convolutional Networks for 3D Human Pose Regressioncited by 204
- 2017Towards Accurate Multi-person Pose Estimation in the Wildcited by 131
- 2020TransPose: Keypoint Localization via Transformercited by 126
- 2019PifPaf: Composite Fields for Human Pose Estimationcited by 120
- 2021Bottom-Up Human Pose Estimation Via Disentangled Keypoint Regressioncited by 112
- 2020End-to-End Object Detection with Adaptive Clustering Transformercited by 109
- 2021TokenPose: Learning Keypoint Tokens for Human Pose Estimationcited by 107
- 2021Pose Recognition with Cascade Transformerscited by 99
- 2019The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimationcited by 79
- 2021Complementary Patch for Weakly Supervised Semantic Segmentationcited by 75
- 2017Multi-Context Attention for Human Pose Estimationcited by 66
- 2017Learning Feature Pyramids for Human Pose Estimationcited by 57
- 2020Differentiable Hierarchical Graph Grouping for Multi-Person Pose Estimationcited by 52
- 2020Inter-Image Communication for Weakly Supervised Localizationcited by 52
- 2017Benchmarking and Error Diagnosis in Multi-Instance Pose Estimationcited by 48
- 2018MultiPoseNet: Fast Multi-Person Pose Estimation using Pose Residual Networkcited by 47
- 2018PoseFix: Model-agnostic General Human Pose Refinement Networkcited by 47
- 2021TFPose: Direct Human Pose Estimation with Transformerscited by 42
- 2018Multi-Scale Structure-Aware Network for Human Pose Estimationcited by 37
- 2021FCPose: Fully Convolutional Multi-Person Pose Estimation with Dynamic Instance-Aware Convolutionscited by 34
- 2020Toward fast and accurate human pose estimation via soft-gated skip connectionscited by 22
- 2020Adversarial Semantic Data Augmentation for Human Pose Estimationcited by 22
- 2020An Empirical Study of the Collapsing Problem in Semi-Supervised 2D Human Pose Estimationcited by 21
- 2018Quantized Densely Connected U-Nets for Efficient Landmark Localizationcited by 20
- 2017Knowledge-Guided Deep Fractal Neural Networks for Human Pose Estimationcited by 18
- 2017Human Pose Estimation using Global and Local Normalizationcited by 10
- 2017Self Adversarial Training for Human Pose Estimationcited by 8
- 2019Cascade Feature Aggregation for Human Pose Estimationcited by 8
Built on by
Citing papers per year: 2023: 1.
1 paper.