The Influence of Faulty Labels in Data Sets on Human Pose Estimation
arXiv:2409.03887 · doi:10.24132/CSRN.2025-2
Abstract
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.
15 pages, 7 figures, 5 tables
References in corpus (22)
- Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models
- ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation
- Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shift
- Cascaded Pyramid Network for Multi-Person Pose Estimation
- Detecting and Correcting for Label Shift with Black Box Predictors
- Pervasive Label Errors in Test Sets Destabilize Machine Learning Benchmarks
- Deep Learning-Based Human Pose Estimation: A Survey
- Human Body Pose Estimation for Gait Identification: A Comprehensive Survey of Datasets and Models
- Distribution-Aware Coordinate Representation for Human Pose Estimation
- Label Noise Types and Their Effects on Deep Learning
- Accounting for Variance in Machine Learning Benchmarks
- Learning Delicate Local Representations for Multi-Person Pose Estimation
- The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimation
- PyOD: A Python Toolbox for Scalable Outlier Detection
- MixSTE: Seq2seq Mixed Spatio-Temporal Encoder for 3D Human Pose Estimation in Video
- SimCC: a Simple Coordinate Classification Perspective for Human Pose Estimation
- Improving Multi-Person Pose Estimation using Label Correction
- Ray3D: ray-based 3D human pose estimation for monocular absolute 3D localization
- P-STMO: Pre-Trained Spatial Temporal Many-to-One Model for 3D Human Pose Estimation
- Detecting Label Errors by using Pre-Trained Language Models
- Bottom-Up 2D Pose Estimation via Dual Anatomical Centers for Small-Scale Persons
- LocLLM: Exploiting Generalizable Human Keypoint Localization via Large Language Model