Search papers, labs, and topics across Lattice.
This paper introduces HuM-Eval, a novel framework for evaluating human motion in generated videos by combining coarse global assessment with fine-grained analysis of anatomical correctness and motion stability. HuM-Eval leverages a Vision Language Model for global quality assessment, 2D pose estimation for anatomical verification, and 3D human motion analysis for stability evaluation. Experiments on a new benchmark, HuM-Bench, demonstrate that HuM-Eval achieves a 58.2% human correlation, surpassing existing metrics and providing a more accurate reflection of human perception.
Current video evaluation metrics miss the mark on human motion quality, but HuM-Eval closes the gap with a new framework that actually aligns with human preferences.
Video generation models have developed rapidly in recent years, where generating natural human motion plays a pivotal role. However, accurately evaluating the quality of generated human motion video remains a significant challenge. Existing evaluation metrics primarily focus on global scene statistics, often overlooking fine-grained human details and consequently failing to align with human subjective preference. To bridge this gap, we propose HuM-Eval, a novel human-centric evaluation framework that adopts a coarse-to-fine strategy. Specifically, our framework first utilizes a Vision Language Model to perform a coarse assessment of global video quality. It then proceeds to a fine-grained analysis, using 2D pose to verify anatomical correctness and 3D human motion to evaluate motion stability. Extensive experiments demonstrate that HuM-Eval achieves an average human correlation of 58.2%, outperforming state-of-the-art baselines. Furthermore, we introduce HuM-Bench, a comprehensive benchmark comprising 1,000 diverse prompts, and conduct a detailed evaluation of existing text-to-video models, paving the way for next-generation human motion generation.
