leofansq / spidermesh

SpiderMesh: Spatial-aware Demand-guided Recursive Meshing for RGB-T Semantic Segmentation (Updating)

Hi there 👋

I'm a Researcher at Institute for AI Industry Research, Tsinghua University (AIR, THU). I obtained my M.S. degree in Automation from Institute of Automation, Chinese Academy of Sciences (CASIA) in 2022, and the B.E. degree in Automation from Shanghai Jiao Tong University (SJTU) in 2019.

• 🔭 I'm currently working on AI for Transportation and AI for Life Science
• 📖 Check my publications via google scholar
• ⚡ Some of my projects are opensourced at GitHub
• Intern Opening (Research Intern @ AIR): Kindly email your resume and availability to [email protected], I will reach back to you if you are a good fit.

Some Tools

Some Researches

spidermesh's Issues

questions about datasets and test performance

Dear Author:

Thank you very much for your open source code. I encountered some problems when reproducing your code, as follows:

1. Your MFNet data set preprocessed by RTFNet shows no permission to download (so I used the original MFNet data set).

2. The best performance I tested after using your train.py is 55.77.

3. After I tested the weight of your resnet152, the result showed that it was only 55.86, which is much lower than the 57.9 in your paper.

Can you enlighten me? Thank you so much!