CVPR 2025 Oral (CCF-A) paper 'Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Models', accepted with full score; ranked 1st in ImageNet 256×256 generation (FID=1.35)
Paper 'ViTMatte: Boosting Image Matting with Pretrained Plain Vision Transformers' published in Information Fusion (IF=18.6 in 2023); integrated into Hugging Face Transformers as a standard matting method and into Nuke, with model weights downloaded over 4 million times monthly
NeurIPS 2024 (CCF-A) paper 'FasterDiT: Towards Faster Diffusion Transformers Training without Architecture Modification'
Paper 'EVA-X: A Foundation Model for General Chest X-ray Analysis with Self-supervised Learning' published in npj Digital Medicine (Nature Partner Journal, IF=15.1)
Multiple first-author or co-first-author papers on topics including mobile deployment of Video-VAEs, large-kernel cell nuclei segmentation, and interactive image matting with Segment Anything Models
China National Scholarship 2024 (awarded to top 0.2% of students nationwide)
Gold Award in China College Students' 'Internet+' Innovation and Entrepreneurship Competition 2022 (success rate: 0.009%)