This dataset contains the benchmark test data for the CVPR 2025 paper ObjectMover (
https://xinyu-andy.github.io/ObjMover/).
ObjMove-A: 200 image sets with objects at two distinct positions per scene and a clean background for reference evaluation.ObjMove-B: 150 image sets from diverse scenes, with complex backgrounds, occlusions, shadows, and manually annotated masks for challenging evaluation.
Results: This contains the results of our method.
