RoboSpatial-Home
A benchmark for evaluating spatial context prediction capability, requiring models to predict object placeable regions in household scenes, with context accuracy as the evaluation metric.
Timeline (last 90 days)
On the RoboSpatial-Home benchmark, Spatial-TRM achieves a context accuracy of 29.51%.
Context accuracy is a metric in the RoboSpatial-Home benchmark that measures the quality of spatial context prediction, typically requiring the IoU between the predicted region and the ground-truth placeable region to exceed a threshold.
All Facts (2)
On the RoboSpatial-Home benchmark, Spatial-TRM achieves a context accuracy of 29.51%.
50%UnverifiedContext accuracy is a metric in the RoboSpatial-Home benchmark that measures the quality of spatial context prediction, typically requiring the IoU between the predicted region and the ground-truth placeable region to exceed a threshold.
50%