Human detectors are surprisingly powerful reward models
This addresses the challenge of generating realistic human motions in videos for applications like sports and dance, though it is incremental as it builds on existing reward optimization methods.
The paper tackled the problem of poor human motion quality in video generation models by proposing HuDA, a simple reward model that uses off-the-shelf detectors to improve motion realism, resulting in a 73% win-rate over state-of-the-art models like Wan 2.1.
Video generation models have recently achieved impressive visual fidelity and temporal coherence. Yet, they continue to struggle with complex, non-rigid motions, especially when synthesizing humans performing dynamic actions such as sports, dance, etc. Generated videos often exhibit missing or extra limbs, distorted poses, or physically implausible actions. In this work, we propose a remarkably simple reward model, HuDA, to quantify and improve the human motion in generated videos. HuDA integrates human detection confidence for appearance quality, and a temporal prompt alignment score to capture motion realism. We show this simple reward function that leverages off-the-shelf models without any additional training, outperforms specialized models finetuned with manually annotated data. Using HuDA for Group Reward Policy Optimization (GRPO) post-training of video models, we significantly enhance video generation, especially when generating complex human motions, outperforming state-of-the-art models like Wan 2.1, with win-rate of 73%. Finally, we demonstrate that HuDA improves generation quality beyond just humans, for instance, significantly improving generation of animal videos and human-object interactions.