WireSeg-36K

WireSeg-36K A 36,000-image dataset for cable / wire instance segmentation: physically simulated deformable linear objects (DLOs) rendered over diverse backgrounds, with exact per-instance masks and per-image metric depth. Single category: cable. Generated with DeformX. Images: 36,000 (1024×1024 RGB) Instances: 176,387 cable masks (COCO RLE) Categories: 1 (cable) Modalities: RGB, instance masks, per-image metric depth Scenes scene family cable counts images… See the full description on the dataset page:

Source: Hugging Face Hub (DeformX/WireSeg-36K). Metadata imported from the dataset’s Hub tags.