The Influence of Faulty Labels in Data Sets on Human Pose Estimation
Not on the map. Why?
Abstract (arXiv)
In this study we provide empirical evidence demonstrating that the quality of training data impacts model performance in Human Pose Estimation (HPE). Inaccurate labels in widely used data sets, ranging from minor errors to severe mislabeling, can negatively influence learning and distort performance metrics. We perform an in-depth analysis of popular HPE data sets to show the extent and nature of label inaccuracies. Our findings suggest that accounting for the impact of faulty labels will facilitate the development of more robust and accurate HPE models for a variety of real-world applications. We show improved performance with cleansed data.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 34 of its 53 references that are in the map. First posted on arXiv 5 Sep 2024.
Builds on
34 of 53 references.
References · 34
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2014Microsoft COCO: Common Objects in Contextcited by 17,498
- 2017Squeeze-and-Excitation Networkscited by 4,763
- 2022Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelscited by 1,970
- 2019Deep High-Resolution Representation Learning for Human Pose Estimationcited by 1,339
- 2018Model Cards for Model Reportingcited by 1,068
- 2016Stacked Hourglass Networks for Human Pose Estimationcited by 913
- 2022ViTPose: Simple Vision Transformer Baselines for Human Pose Estimationcited by 450
- 2020ResNeSt: Split-Attention Networkscited by 445
- 20183D human pose estimation in video with temporal convolutions and semi-supervised trainingcited by 401
- 2021Pervasive Label Errors in Test Sets Destabilize Machine Learning Benchmarkscited by 388
- 2018Detecting and Correcting for Label Shift with Black Box Predictorscited by 364
22 more
- 2017Cascaded Pyramid Network for Multi-Person Pose Estimationcited by 341
- 2018Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shiftcited by 240
- 2019PyOD: A Python Toolbox for Scalable Outlier Detectioncited by 221
- 2020Deep Learning-Based Human Pose Estimation: A Surveycited by 205
- 2021Accounting for Variance in Machine Learning Benchmarkscited by 195
- 2019Distribution-Aware Coordinate Representation for Human Pose Estimationcited by 165
- 2022MixSTE: Seq2seq Mixed Spatio-Temporal Encoder for 3D Human Pose Estimation in Videocited by 158
- 2020BlazePose: On-device Real-time Body Pose trackingcited by 112
- 2022Recovering 3D Human Mesh from Monocular Images: A Surveycited by 103
- 2022P-STMO: Pre-Trained Spatial Temporal Many-to-One Model for 3D Human Pose Estimationcited by 80
- 2019The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimationcited by 79
- 2021SimCC: a Simple Coordinate Classification Perspective for Human Pose Estimationcited by 68
- 2020Learning Delicate Local Representations for Multi-Person Pose Estimationcited by 65
- 2017Benchmarking and Error Diagnosis in Multi-Instance Pose Estimationcited by 48
- 2020Label Noise Types and Their Effects on Deep Learningcited by 29
- 2022Ray3D: ray-based 3D human pose estimation for monocular absolute 3D localizationcited by 25
- 20222D Human Pose Estimation: A Surveycited by 21
- 2022Detecting Label Errors by using Pre-Trained Language Modelscited by 16
- 2022Bottom-Up 2D Pose Estimation via Dual Anatomical Centers for Small-Scale Personscited by 5
- 2023Human Body Pose Estimation for Gait Identification: A Comprehensive Survey of Datasets and Modelscited by 5
- 2018Improving Multi-Person Pose Estimation using Label Correctioncited by 2
- 2022Machine Learning State-of-the-Art with Uncertaintiescited by 1
Built on by
Citing papers per year: 2025: 1; 2026: 1.
2 papers.