facebookresearch/CutLER

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1072 stars · 109 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

CutLER is a research codebase from Meta FAIR for training object detection and instance segmentation models without human annotations, using the MaskCut pseudo-mask approach on ImageNet-1K. It also includes VideoCutLER for unsupervised video instance segmentation without optical flow.

Use cases

• train an object detector without labeled data
• generate pseudo-masks for objects in images
• unsupervised instance segmentation
• unsupervised video instance segmentation
• pretrain a detector for semi-supervised fine-tuning
• evaluate detection robustness across domain shifts

When to choose

• you lack annotated detection/segmentation data
• you need a pretrained backbone for semi-supervised detection
• you want to reproduce CVPR research on unsupervised segmentation

When to avoid

• you need a production-ready, supported detection framework
• you have abundant labeled data and just need standard supervised training
• you need a no-GPU or lightweight inference tool

Facets

library · maturity active

machine-learning computer-vision image-processing video-processing computer-vision machine-learning deep-learning artificial-intelligence python object-detection instance-segmentation unsupervised-learning pseudo-labeling research-code pytorch cvpr linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebookresearch/CutLER")