Event
SpatialScout Active Spatial Understanding Benchmark Released
1 reports · 1 sources
The SpatialScout author team has released SpatialScout, an active spatial understanding benchmark, along with its accompanying dataset and code. Unlike given viewpoints or fixed routes, this benchmark hands the camera to the AI, letting the model autonomously decide when to turn, move, and take photos in previously unseen virtual rooms, and submit the 3D coordinates of the target object. The official evaluation selects 80 indoor scenes and 3,140 target instances, with scenes drawn from iTHOR, RoboTHOR, ArchitecTHOR, ReplicaCAD, and ProcTHOR; each task allows at most 100 rounds of model interaction. Scoring considers both the average IoU and the proportion of IoU exceeding 0%, 50%, and 90%. Costs are estimated based on a fixed OpenRouter price snapshot, averaging only over tasks with complete token records, and do not represent actual billing.
Report timeline
- SpatialScout benchmark: enabling agents to actively observe and build 3D spatial cognition
Xinzhiyuan reported on the SpatialScout benchmark, which gives a camera to an AI and places it in previously unseen virtual rooms, where the model autonomously decides when to turn, move, and capture images, and submits the 3D coordinates of target objects. The formal evaluation selects 80 indoor scenes and 3,140 target instances from iTHOR, RoboTHOR, ArchitecTHOR, ReplicaCAD, and ProcTHOR, with at most 100 interaction rounds per task; scoring considers both mean IoU and the proportion of IoU above 0%, 50%, and 90%.
新智元 · 微信公众号 ·