sam3 — Meta Segment Anything Model 3
Meta’s SAM 3 segmentation model — promptable image and video segmentation. The image bakes in the facebook/sam3 checkpoint at build time so cold-start is instant.
Image tag
docker.io/manvarharsh/sam3:cuda12
What’s in this image
- Base:
nvidia/cuda:12.6.3-cudnn-devel-ubuntu22.04 - Python 3.12 (conda)
- PyTorch 2.10 (CUDA 12.8 wheels)
- SAM 3 (cloned from
facebookresearch/sam3) - Bundled checkpoint at
/opt/sam3_checkpoints/sam3.pt - Optional flash-attn-3
- JupyterHub with Podstack authenticator
- OpenSSH server
Default ports
| Port | Service |
|---|---|
| 22 | SSH |
| 8000 | JupyterHub |
Use cases
- Promptable segmentation (point / box / mask prompts)
- Video object tracking / segmentation
- Building annotation tools and dataset prep pipelines
- Research on top of SAM 3 weights
Environment variables
| Variable | Description |
|---|---|
ENABLE_SSH | Enable SSH server |
ENABLE_JUPYTERHUB | Enable JupyterHub on port 8000 |
PODSTACK_API_URL | Backend URL for JupyterHub token validation |
SAM3_CHECKPOINT_DIR | Override checkpoint directory (default /opt/sam3_checkpoints) |
SAM3_CHECKPOINT | Override specific checkpoint file (default sam3.pt; use e.g. sam3.1_multiplex.pt) |
SSH_PUBLIC_KEY | Public key for SSH |
Persistence
Mount at /data. Place input images / videos in /data/inputs/ and write masks / annotations to /data/output/. Point SAM3_CHECKPOINT_DIR at /data/checkpoints/ if you want to use a custom checkpoint stored on the volume.