Jiachen Fu(付嘉晨)

Jiachen Fu(付嘉晨)

LLMs, VLMs, Multimodal AI, Trustworthy AI

Publications

You can also find my articles on my Google Scholar profile.

Journal Articles

ViDA-UGC: Detailed Image Quality Analysis via Visual Distortion Assessment for UGC Images

Published in arXiv preprint, 2025

ViDA-UGC proposes a detailed image quality analysis framework for User-Generated Content (UGC) images via visual distortion assessment, providing fine-grained quality evaluation across diverse distortion types.

Recommended citation: Liao, W., Yuan, J., Xu, Y., Guo, C., Zhang, Z., Li, J., Fu, J., et al. (2025). ViDA-UGC: Detailed Image Quality Analysis via Visual Distortion Assessment for UGC Images. arXiv preprint arXiv:2508.12605.
Download Paper

Conference Papers

DetectAnyLLM: Towards Generalizable and Robust Detection of Machine-Generated Text Across Domains and Models

Published in ACM International Conference on Multimedia (ACM MM 2025, CCF-A), 2025

We propose DetectAnyLLM, a novel and robust approach for detecting machine-generated text across diverse domains and LLMs. Our optimization strategy reduces computation cost while achieving ~70% relative improvement over previous SOTA methods, along with a comprehensive benchmark dataset for robust evaluation.

Recommended citation: Fu, J., Guo, C., & Li, C. (2025). DetectAnyLLM: Towards Generalizable and Robust Detection of Machine-Generated Text Across Domains and Models. ACM MM 2025. Oral.
Download Paper

EvalMuse-40K: A Reliable and Fine-Grained Benchmark with Comprehensive Human Annotations for Text-to-Image Generation Model Evaluation

Published in AAAI 2026 (CCF-A), 2024

We present EvalMuse-40K, a large-scale fine-grained benchmark with comprehensive human annotations for evaluating text-to-image generation models. We fine-tune BLIP to produce human-aligned scores with fine-grained capability, and use the dataset to fairly rank 20+ T2I models.

Recommended citation: Han, S., Fan, H., Fu, J., Li, L., Li, T., et al. (2024). EvalMuse-40K: A Reliable and Fine-Grained Benchmark with Comprehensive Human Annotations for Text-to-Image Generation Model Evaluation. AAAI 2026.
Download Paper