Yuxiang Lin

lin.yuxiang.contact@gmail.com

I received my MS ECE degree from the Georgia Institute of Technology, where my research interests centered on Multimodal Large Language Models (MLLM) and Affective Computing. Previously, I gained industrial algorithm experience as an intern at Tencent and Baidu.

My current interests include:

AI Application & Memory Systems
Graph-based RAG Systems
Multi-agent Collaboration
Large Language Models (LLMs) & Multimodal LLMs

I am also an open source programmer. I created Understand Anything — a Claude Code plugin that turns any codebase into an interactive knowledge graph you can explore, search, and ask questions about. It combines LLM intelligence with static analysis to produce a living, explorable map of your project with plain-English explanations for everything.

Recently, we released research-git — an Agentic Git that captures a code idea as a clean semantic unit and regenerates it onto today’s codebase. Feel free to give it a try!

Open to opportunities! I am happy to discuss any collaboration or role — feel free to reach out.

news

Jul 05, 2026	We recently released research-git — capture a code idea as a clean semantic unit and regenerate it onto today’s codebase, aka Agentic Git. Feel free to give it a try!
Jun 27, 2026	Invited by Apache Flink and Alibaba Cloud to give a talk at Flink Forward Asia 2026 in Shenzhen — check out the LinkedIn post.
Jun 09, 2026	Three months after launch, Understand Anything hit 55,000 GitHub stars on June 9, growing from 0 to 55k at a remarkable pace.
Mar 20, 2026	Understand Anything reached 1,000 GitHub stars in just 5 days after launch. Deeply grateful to the open-source community for the incredible support!
Jul 01, 2025	Try MER-Factory for automatic construction of multimodal emotion recognition and reasoning datasets.
Jun 01, 2025	One paper about Benchmarking VLLM’s Emotion Interpretation ability is accepted by CVPR Workshop, NeXD (1/3 Oral).
May 01, 2025	Started my internship at Tencent.
Dec 01, 2024	One paper about Multimodal Large Language Model in Emotion Reasoning is accepted by NeurIPS (CCF-A).
Jul 01, 2024	One co-first author paper about invisible gas detection is accepted by CVIU (JCR Q1, CCF-B).
Mar 01, 2024	One paper about Conversational Emotion-Cause Pair Analysis with LLM is accepted by SemEval 2024, NAACL.
Jan 15, 2024	Started my internship at Baidu Inc.
Jan 01, 2024	Awarded the First Prize of Research and Innovation Award (3000 CNY) and Star of Craftsmanship (3000 CNY).
Aug 01, 2023	My instance segmentation tutorial has been featured in MMYOLO v0.6.0 highlight! Check out the tutorial to master the essentials of instance segmentation.
Jul 15, 2023	One paper on multimodal emotion recognition is accepted by ACM MM 2023!
Jul 01, 2023	We are the runner-up in the Grand Challenge (MER 2023) of ACM MM!

selected publications

ArXiv

Why We Feel: Breaking Boundaries in Emotional Reasoning with Multimodal Large Language Models

Yuxiang Lin, Jingdong Sun, Zhi-Qi Cheng, and 7 more authors

arXiv preprint arXiv:2504.07521, 2025

Abs HTML PDF Code

This paper presents EIBench, a comprehensive benchmark for evaluating Vision-Language Large Models’ ability to interpret and reason about emotions across multiple modalities.
NeurIPS

Emotion-LLaMA: Multimodal Emotion Recognition and Reasoning with Instruction Tuning

Zebang Cheng, Zhi-Qi Cheng, Jun-Yan He, and 5 more authors

In Advances in Neural Information Processing Systems (NeurIPS), 2024

Abs PDF Code

This paper introduces Emotion-LLaMA, a multimodal large language model for emotion recognition and reasoning with instruction tuning. The model achieves state-of-the-art performance on multiple emotion recognition benchmarks.
CVIU

Invisible Gas Detection: An RGB-Thermal Cross Attention Network and A New Benchmark

Jue Wang, Yuxiang Lin, Qi Zhao, and 4 more authors

Computer Vision and Image Understanding, 2024

Abs HTML PDF Code

This paper proposes an RGB-Thermal Cross Attention Network (RT-CAN) for invisible gas detection and introduces a new benchmark dataset for this task.