🚀 Journal & Conference Acceptances
- 04/2025 – Nature Communication: GrAIL has 1 paper accepted to Nature Communication.
- 03/2025 – CVPR 2025: GrAIL has 41 papers accepted to CVPR 2025.
- 01/2025 – ICLR 2025: GrAIL has 25 papers accepted to ICLR 2025.
- 01/2025 – NAACL 2025: GrAIL has 7 papers accepted to NAACL 2025.
- 12/2024 – AAAI 2025: GrAIL has 10 papers accepted to AAAI 2025.
- 11/2024 – COLING 2025: GrAIL has 6 papers accepted to COLING 2025.
🌟 Featured Research
Chen Change Loy’s work MEAT, by integrating mesh attention and keypoint conditioning, significantly reduces the complexity of multi-view attention at 1024×1024 resolution while preserving cross-view consistency. Moreover, by leveraging multi-view human motion videos for training, it effectively produces dense, consistent human images that outperform existing multi-view diffusion methods.
🔗 https://johann.wang/MEAT/
Chen Change Loy’s work F-LMM freezes off-the-shelf LMMs, uses minimal CNN layers plus a SAM-based mask refiner to translate word-pixel attention into segmentation masks, and thereby preserves the original conversational ability while delivering robust visual grounding and segmentation performance.
🔗 https://github.com/wusize/F-LMM
Assistant Professor Xingang Pan’s work DragGAN, a revolutionary interactive image editing tool, has been widely reported by media outlets including The Independent, Business Insider, The Verge, etc. Its open-sourced code has received over 35k GitHub stars.
🎓 Student Recognition
PhD student Zhaoxi Chen (under Assoc Prof. Liu Ziwei) and Yushi Lan (under Prof. Chen Change Loy) have been awarded the Meshy Fellowship.
🔗 https://www.meshy.ai/blog/fellowship-2025
