MRQ Dataset

MRQ Dataset MRQ is a large-scale synthetic computer vision dataset with rendered multi-scene trajectories. It provides paired RGB frames, depth maps, optical-flow encodings, instance labels, camera poses, object/id mappings, and tracking annotations for depth, flow, segmentation, pose, and synthetic-to-real vision research. Dataset Composition The local source dataset is approximately 4.8 TB. The current Hub release is generated by the scene-packing upload… See the full description on the dataset page:

Source: Hugging Face Hub (Yukki1011/PromptDepth). Metadata imported from the dataset’s Hub tags.