-
[pdf]
[arXiv]
[bibtex]@InProceedings{Xu_2025_ICCV, author = {Xu, Peng and Xiong, Shengwu and Zhang, Jiajun and Chen, Yaxiong and Zhou, Bowen and Loy, Chen Change and Clifton, David and Lee, Kyoung Mu and Van Gool, Luc and He, Ruiming and Yao, Ruilin and Long, Xinwei and Huang, Jirui and Tian, Kai and Yang, Sa and Shao, Yihua and Feng, Jin and Zhong, Yue and Zhou, Jiakai and Tang, Cheng and Zou, Tianyu and Zhang, Yifang and Liang, Junming and Li, Guoyou and Wang, Zhaoxiang and Zhou, Qiang and Zhao, Yichen and Xiong, Shili and Nam, Hyeongjin and Lee, Jaerin and Chung, Jaeyoung and Park, JoonKyu and Oh, Junghun and Lee, Kanggeon and Lee, Wooseok and Ro, Juneyoung and Osman, Turghun and Hu, Can and Liao, Chaoyang and Chen, Cheng and Han, Chengcheng and Qiu, Chenhao and Peng, Chong and Xu, Cong and Li, Dailin and Wang, Feiyu and Gao, Feng and Zhu, Guibo and Tang, Guopeng and Lu, Haibo and Fang, Han and Qi, Han and Wu, Hanxiao and Cheng, Haobo and Sun, Hongbo and Chen, Hongyao and Hu, Huayong and Li, Hui and Ma, Jiaheng and Yu, Jiang and Wang, Jianing and Yang, Jie and He, Jing and Zhou, Jinglin and Li, Jingxuan and Kittler, Josef and Zheng, Lihao and Zhao, Linnan and Jia, Mengxi and Yan, Muyang and Nguyen, Thanh Thien and Luo, Pu and Li, Qi and Song, Shien and Dong, Shijie and Shao, Shuai and Li, Shutao and Xue, Taofeng and Xu, Tianyang and Gao, Tianyi and Li, Tingting and Zhang, Wei and Su, Weiyang and Dong, Xiaodong and Wu, Xiao-Jun and Zhou, Xiaopeng and Chen, Xin and Wei, Xin and You, Xinyi and Kang, Xudong and Zhou, Xujie and Liu, Xusheng and Wang, Yanan and Huang, Yanbin and Liu, Yang and Yang, Yang and Deng, Yanglin and Kang, Yashu and Yuan, Ye and Wen, Yi and Tian, Yicen and Tao, Yilin and Tang, Yin and Lin, Yipeng and Wang, Yiqing and Xi, Yiting and Yu, Yongkang and Li, Yumei and Qin, Yuxin and Chen, Yuying and Cen, Yuzhe and Zou, Zhaofan and Liu, Zhaohong and Shen, Zhehao and Du, Zhenglin and Li, Zhengyang and Huang, Zhenni and Shao, Zhenwei and Song, Zhilong and Feng, Zhiyong and Wang, Zhiyu and Yu, Zhou and Li, Ziang and Zhai, Zihan and Zhang, Zijian and Peng, Ziyang and Xiao, Ziyun and Li, Zongshu}, title = {MARS2 2025 Challenge on Multimodal Reasoning: Datasets, Methods, Results, Discussion, and Outlook}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops}, month = {October}, year = {2025}, pages = {6517-6546} }
MARS2 2025 Challenge on Multimodal Reasoning: Datasets, Methods, Results, Discussion, and Outlook
Abstract
This paper reviews the MARS2 2025 Challenge on Multimodal Reasoning. We aim to bring together different approaches in multimodal machine learning and LLMs via a large benchmark. We hope it better allows researchers to follow the state-of-the-art in this very dynamic area. Meanwhile, a growing number of testbeds have boosted the evolution of general-purpose large language models. Thus, this year's MARS2 focuses on real-world and specialized scenarios to broaden the multimodal reasoning applications of MLLMs. Our organizing team released two tailored datasets Lens and AdsQA as test sets, which support general reasoning in 12 daily scenarios and domain-specific reasoning in advertisement videos, respectively. We evaluated 40+ baselines that include both generalist MLLMs and task-specific models, and opened up three competition tracks, i.e., Visual Grounding in Real-world Scenarios (VGRS), Visual Question Answering with Spatial Awareness (VQA-SA), and Visual Reasoning in Creative Advertisement Videos (VR-Ads). Finally, 76 teams from the renowned academic and industrial institutions have registered and 40+ valid submissions (out of 1200+) have been included in our ranking lists. Our datasets, code sets (40+ baselines and 15+ participants' methods), and rankings are publicly available on the MARS2 workshop website and our GitHub organization page https://github.com/mars2workshop/, where our updates and announcements of upcoming events will be continuously provided.
Related Material