Fine-Grained Human Pose Editing Assessment via Layer-Selective MLLMs

TL;DR

This paper introduces HPE-Bench, a comprehensive benchmark and a layer-selective MLLM framework for fine-grained evaluation of human pose editing, addressing current limitations in authenticity and quality assessment.

Contribution

It presents a new benchmark with standardized samples and a novel layer-sensitive MLLM framework for improved pose editing evaluation.

Findings

HPE-Bench contains 1,700 samples from 17 models.

The proposed framework outperforms existing methods in authenticity detection.

Layer sensitivity analysis enhances pose evaluation accuracy.

Abstract

Text-guided human pose editing has gained significant traction in AIGC applications. However,it remains plagued by structural anomalies and generative artifacts. Existing evaluation metrics often isolate authenticity detection from quality assessment, failing to provide fine-grained insights into pose-specific inconsistencies. To address these limitations, we introduce HPE-Bench, a specialized benchmark comprising 1,700 standardized samples from 17 state-of-the-art editing models, offering both authenticity labels and multi-dimensional quality scores. Furthermore, we propose a unified framework based on layer-selective multimodal large language models (MLLMs). By employing contrastive LoRA tuning and a novel layer sensitivity analysis (LSA) mechanism, we identify the optimal feature layer for pose evaluation. Our framework achieves superior performance in both authenticity detection and multi-dimensional quality regression, effectively bridging the gap between forensic detection and quality assessment.