Test-time Scaling for Computer Vision


On the Suitability of Reinforcement Fine-Tuning to Visual Tasks
Xiaxu Chen,
Wei Li,
Chunxu Liu,
Chi Xie,
Xiaoyan Hu,
Chengqian Ma,
Feng Zhu,
Rui Zhao
[pdf]
[bibtex]
@InProceedings{Chen_2025_CVPR, author = {Chen, Xiaxu and Li, Wei and Liu, Chunxu and Xie, Chi and Hu, Xiaoyan and Ma, Chengqian and Zhu, Feng and Zhao, Rui}, title = {On the Suitability of Reinforcement Fine-Tuning to Visual Tasks}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {3382-3386} }

Get a GRIP on Test Time Adaptation! - Group Robust Inference-Time Policy Optimization for Vision Models
Prabhav Sanga,
Jaskaran Singh,
Tapabrata Chakraborti
[pdf]
[bibtex]
@InProceedings{Sanga_2025_CVPR, author = {Sanga, Prabhav and Singh, Jaskaran and Chakraborti, Tapabrata}, title = {Get a GRIP on Test Time Adaptation! - Group Robust Inference-Time Policy Optimization for Vision Models}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {3401-3410} }

Scaling Test-Time Compute Can Outperform Larger Architectures in Computer Vision
Erfan Darzi,
Dylan Nguyen,
George Cheng
[pdf]
[bibtex]
@InProceedings{Darzi_2025_CVPR, author = {Darzi, Erfan and Nguyen, Dylan and Cheng, George}, title = {Scaling Test-Time Compute Can Outperform Larger Architectures in Computer Vision}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {3392-3400} }

TTGen: Incorporating Test-time Scaling to Diffusion Models
Yuming Qiao,
Yuechen Wang,
Xudong Zhang,
Dan Meng
[pdf]
[bibtex]
@InProceedings{Qiao_2025_CVPR, author = {Qiao, Yuming and Wang, Yuechen and Zhang, Xudong and Meng, Dan}, title = {TTGen: Incorporating Test-time Scaling to Diffusion Models}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops}, month = {June}, year = {2025}, pages = {3387-3391} }