PoseAdapt: Sustainable Human Pose Estimation via Continual Learning Benchmarks and Toolkit
This work provides a crucial toolkit and benchmarks for researchers and practitioners in computer vision, enabling more sustainable and adaptable human pose estimation models by reducing the need for costly full retraining.
The paper introduces PoseAdapt, an open-source framework and benchmark suite for continual human pose estimation model adaptation. It addresses the inefficiency of retraining models from scratch by providing domain-incremental and class-incremental tracks that simulate real-world changes in data, and allows for benchmarking continual learning methods and adapting pretrained models with minimal supervision.
Human pose estimators are typically retrained from scratch or naively fine-tuned whenever keypoint sets, sensing modalities, or deployment domains change--an inefficient, compute-intensive practice that rarely matches field constraints. We present PoseAdapt, an open-source framework and benchmark suite for continual pose model adaptation. PoseAdapt defines domain-incremental and class-incremental tracks that simulate realistic changes in density, lighting, and sensing modality, as well as skeleton growth. The toolkit supports two workflows: (i) Strategy Benchmarking, which lets researchers implement continual learning (CL) methods as plugins and evaluate them under standardized protocols; and (ii) Model Adaptation, which allows practitioners to adapt strong pretrained models to new tasks with minimal supervision. We evaluate representative regularization-based methods in single-step and sequential settings. Benchmarks enforce a fixed lightweight backbone, no access to past data, and tight per-step budgets. This isolates adaptation strategy effects, highlighting the difficulty of maintaining accuracy under strict resource limits. PoseAdapt connects modern CL techniques with practical pose estimation needs, enabling adaptable models that improve over time without repeated full retraining.