facebookresearch/DensePose

Health v2 · maintenance only

Flags: no_releases archived no_license

Adoption not part of the score

7259 stars · 1327 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

DensePose is a research library from Facebook AI that maps all human pixels in 2D RGB images to a 3D surface-based model of the human body in real time. It includes DensePose-RCNN (built on Detectron/Caffe2), the DensePose-COCO dataset, and notebooks for visualization and texture transfer.

Use cases

• estimate dense human pose from rgb images
• map 2d human pixels to 3d body surface
• train a densepose-rcnn model on coco
• visualize densepose-coco annotations on the smpl model
• transfer textures onto a 3d human body model
• run real-time human body surface estimation

When to choose

• you need the original DensePose-RCNN implementation or the DensePose-COCO dataset for research
• you want to reproduce results from the DensePose paper
• you need image-to-SMPL surface correspondence annotations

When to avoid

• you want maintained code or new architectures - use the DensePose project inside Detectron2 instead
• you need production pose estimation without deep learning infrastructure
• you cannot set up Caffe2/Detectron, which are also deprecated

Facets

library · maturity abandoned

computer-vision machine-learning deep-learning image-processing computer-vision machine-learning artificial-intelligence python pose-estimation human-body-modeling detectron caffe2 research-code dataset smpl linux gpu

Member repositories

For agents

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