Xiamen University and Nanyang Technological University propose RoTA-Bench 3D pose benchmark
What happened
A team from Xiamen University and Nanyang Technological University proposed the RoTA-Bench benchmark, which uses interactive 3D-to-2D projection alignment to evaluate whether vision-language models can precisely adjust an object's 3D pose based on visual feedback. The model must choose among 24 discrete rotation actions, and the task is considered successful only when the intersection-over-union (IoU) between the current projection and the target silhouette reaches 0.90; evaluation allows at most 48 decisions and includes active-stop and stagnation-termination rules. Researchers restricted the interaction to pure rotation to separate 3D orientation reasoning from factors such as translation, grasping, and contact friction, focusing the problem on "how to control pose based on visual evidence." RoTA-Bench contains a total of 674 tasks, of which 100 form a frozen test set and the remaining 574 are reserved for subsequent training and research use.
AI 根据报道生成 · 5 days ago更新
Reporting timeline
Follow the reporting to explore different perspectives.
- 新智元 · 微信公众号Xiamen University and NTU propose RoTA-Bench: interactive projection alignment benchmark for 3D pose reasoning
A team from Xiamen University and Nanyang Technological University proposed RoTA-Bench, which uses interactive 3D-to-2D projection alignment to evaluate whether vision-language models can precisely adjust an object's 3D pose based on visual feedback. The benchmark contains 674 tasks, of which 100 form a frozen test set; models must choose among 24 discrete rotation actions so that the projection reaches an IoU of 0.90 with the target contour.
Activity for this event
- Current
- 3
- Peak
- 7Nov 6 00:00
- Past 24 hours
- -50%
The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.