Personal Homepage:https://sites.google.com/site/poppinace/

Academic Focus: "Foundation Models + Cross-Disciplinary Applications"

My research centers onvisual dense prediction- fundamental computer vision tasks involving pixel-wise labeling, including:

• Object detection

• Semantic/instance segmentation

• Depth estimation

• Image matting

• Image denoising

• Image reconstruction

Recent Research Directions:

1. Design and optimization of universal components

2. Task-specific model architecture development

3. Customized models for "AI+Smart Agriculture" applications

   
   

Representative Achievements:

✓Dynamic upsampling operators:

• IndexNet, SAPA, FADE, A2U, DySample (task-agnostic feature upsampling solutions)

  ✓IndexNet Matting:

• The earliest open-source deep image matting model (ranked #2 in citations among deep matting papers per Google Scholar)

• Widely adopted as benchmark by researchers

  ✓TasselNet series:

• Regression-based local counting models for plant phenotyping

• Internationally recognized in plant science communities

Research Philosophy:

I pursue fundamental research that delivers simple, effective, and generalizable solutions with real-world impact.

Current Grants:

• PI: NSFC Young Scientists Fund (2022.1-2024.12, ongoing)

• Key Member:

• NSFC General Program (2019.1-2022.12, completed)

• MOST Key R&D Program Subproject (2022.10-2025.9, ongoing)