Peilin Feng

Peilin Feng

Beijing University of Aeronautics and Astronautics

My name is Peilin Feng (冯沛林), an undergraduate student at Beihang University(2021-2025), majoring in Automation Science and Electrical Engineering with a minor in Mathematical Sciences. Under the supervision of Professor Lei Huang, I am currently focusing on Deep neural network architecture design and Multimodal Understanding. In the future, I hope to work on vllm and agent. My ultimate goal is to make AI better serve our lives.

I also have a deep passion for mathematics. I serve as a teaching assistant for the Mathematical Analysis for Engineering course. My responsibilities include grading homework, answering students’ questions, and leading review lectures. Outside of my academic and research interests, I enjoy playing ping pong 🏓 and swimming 🤿. In my leisure time, I like to listen to some music 🎵 and watch my favorite cartoon Doraemon 📺 for relax.

My research interests span the intersection of NLP and Computer Vision, with a particular focus on Multimodal Understanding. This stems from my dedication to creating an AGI similar to Doraemon 🐱. Currently, I am actively working to bridge my knowledge gaps in this area, with the goal of becoming an expert in the field.

However, artificial intelligence is developing rapidly, so I always hope to be at the forefront of the field and use my intelligence to devote to our society. I believe that is an exciting exploring travel.

🔥 News

  • 2025.09:  🎉🎉 We released Earth-Agent (First Author), the first agent in the field of EO that can perform multi-step inference tasks on spectral and RGB data.
  • 2025.09:  🎉🎉 Our paper Spot the Fake (Second Author) was accepted at NeurIPS (Poster).
  • 2025.06:  🎓🎓 I reveived my bachelor’s degree at Beihang University, thanks to my teachers and friends, miss you!
  • 2025.06:  🎉🎉 Our paper LEGION was accepted at ICCV (Highlight).
  • 2025.01:  😊😊 I am excited to gain the opportunity to intern with the Multimodal Group at the Shanghai National AI Lab from Feb. 2025. I look forward to collaborating with like-minded mentors and peers on interesting and meaningful work.
  • 2024.12:  😭😭 My first paper to ICLR 2025 was unfortunately rejected. While it’s true that most people’s first submission is full of ups and downs, I still feel quite heartbroken. I hope my revised version will stand out at ICML. Fighting💪!
  • 2024.10:  🎉🎉 I successfully submitted my first paper to ICLR 2025 (First Author), marking a milestone to my research career. I hope I will be more excellent in the future.

📝 Publications

sym

Earth-Agent: Unlocking the Full Landscape of Earth Observation with Agents

Peilin Feng†, Zhutao Lv†, Junyan Ye, Xiaolei Wang, Xinjie Huo, Jinhua Yu, Wanghan Wu, Wenlong Zhang, Lei Bai, Conghui He, Weijia Li✉

💡 Contribution Project Dataset GitHub Stars Report

  • We propose Earth-Agent, a revolutionary paradigm shift from traditional MLLMs to agentic EO analysis, unifying RGB and spectral EO data within an MCP-based tool ecosystem
  • In order to comprehensivly evaluate Earth-Agent, we propose Earth-Bench, which covers Spectrum, Products and RGB modality for scientific workflows requring tool interaction,
  • Earth-Agent substantially outperforms general agents and surpasses remote sensing MLLMs on remote sensing benchmarks, demonstrating both effectiveness and potential for advancing EO research
NeurIPS 2025
sym

Spot the Fake: Large Multimodal Model-Based Synthetic Image Detection with Artifact Explanation

Siwei Wen†, Junyan Ye†, Peilin Feng, Hengrui Kang, Zichen Wen, Yize Chen, Jiang Wu, Wenjun Wu, Conghui He, Weijia Li✉

💡 Contribution Dataset GitHub Stars

  • Introduced FakeClue, a comprehensive dataset containing over 100,000 synthetic and real images across seven categories, annotated with fine-grained artifact clues in natural language.
  • Proposed FakeVLM, a specialized large multimodal model for synthetic image and DeepFake detection, combining authenticity classification with natural language explanations of image artifacts.
ICCV Highlight
sym

LEGION: Learning to Ground and Explain for Synthetic Image Detection

Hengrui Kang†, Siwei Wen†, Zichen Wen†, Junyan Ye, Weijia Li✉, Peilin Feng, Baichuan Zhou, Bin Wang, Dahua Lin, Linfeng Zhang, Conghui He✉

💡 Contribution Dataset GitHub Stars

  • Introduced SynthScars, a high-quality synthetic image detection dataset featuring diverse content types, fine-grained pixel-level artifact annotations, and detailed textual explanations.
  • Proposed LEGION: a multimodal large language model-based framework for artifact localization, explanation generation, and forgery detection, enhancing interpretability in synthetic image analysis.
sym

Layer-Wise Analysis in Exploring the Normalization Strategies in Mamba

Peilin Feng, Yuanshuai Wang, Yunhao Ni, Junle Wang, zhangkui, Wenjun Wu, Lei Huang✉

💡 Contribution

  • We track the spectral norms and singular values of covariance matrices in Mamba to demonstrate the critical role of Norm2 in stability.
  • We track the condition number of K-FAC in Mamba to demonstrate the critical role of Norm1 in improving optimization.
  • We proposed a composite normalization strategy and evaluate its effectiveness across diverse tasks,including sequence modeling, NLP & CV
sym

Bench-CoE: a Framework for Collaboration of Experts from Benchmark

Yuanshuai Wang†, Xingjian Zhang†, Jinkun Zhao†, Siwei Wen, Peilin Feng, Shuhao Liao, Lei Huang✉, Wenjun Wu✉

💡 Contribution GitHub Stars

  • We proposed a novel framework for Bench-CoE that relies solely on the Benchmark when training the router.
  • We comprehensively evaluated the performance and generalization ability of our Bench-CoE model across three stages: Naive, In-distribution, and Out-of-distribution.

🎖 Honors and Awards

  • 2025.06 Outstanding undergraduate graduate of Beihang University(top3%)
  • 2024.12 Grand prize for Contests Scholarship in Beihang University
  • 2023.12 Outstanding student cadres
  • 2023.12 First Prize in the Chinese Mathematics Competition

  • 2023.12 First Prize in Beijing Mathematics Competition (114th among 20000+ students)

  • 2023.12 Second Prize for Contests Scholarship in Beihang University
  • 2023.12 Second Prize in China Undergraduate Mathematical Contest in Modeling
  • 2023.09 Outstanding Teaching Assistant in Mathematical Analysis(6/32)
  • 2023.06 Second Prize in Qixian Cup Robot Challenge Competition(4/32)
  • 2022.12 First Prize Scholarship in Beihang University (top5%)
  • 2022.06 Second Prize in Feng Ru Cup Science and Technology Innovation Competition

📖 Educations

  • 2022.09 - 2025.06 (Expected), Mathematical Science (Minor), Beihang University, Beijing, China.
  • 2021.09 - 2025.06 (Expected), Automation Science (Major), Beihang University, Beijing, China.

💬 Tutorial Materials in Teaching Assistant

🎤 Lectures:

  • 2022.11,  Tutorial, Mathematical Analysis: Indefinite Integral. | [video]
  • 2022.10,  Tutorial, Mathematical Analysis: Equivalent Transform. | [video]
  • 2022.09,  Tutorial, Mathematical Analysis: Limit in Recursion sequence. | [video]
  • 2021.12,  Tutorial, Mathematical Analysis: Integration by parts. | [video]
  • 2021.11,  Tutorial, Mathematical Analysis: Rolle’s theorem. | [video]

📄 Test:

  • 2022.10,  Mock Mid-term Exam, Mathematical Analysis(I) | [Test] & | [Answer]
  • 2023.04,  Mock Mid-term Exam, Mathematical Analysis(II) | [Test] & | [Answer]

💻 Internships

SHLAB
CAST

✏️ Community Service

  • Conference Reviewer: AAAI