# Use SAM2 with videos
import torch
from sam2.sam2_video_predictor import SAM2VideoPredictor
predictor = SAM2VideoPredictor.from_pretrained(mfranzon/sam2-tiny-tracking)
with torch.inference_mode(), torch.autocast("cuda", dtype=torch.bfloat16):
state = predictor.init_state(<your_video>)
# add new prompts and instantly get the output on the same frame
frame_idx, object_ids, masks = predictor.add_new_points(state, <your_prompts>):
# propagate the prompts to get masklets throughout the video
for frame_idx, object_ids, masks in predictor.propagate_in_video(state):
...SAM2 Tiny (Cell and Bacterium Tracking)
The SAM2 tiny checkpoint used in the Model Garden Cell and Bacterium Tracking demo: class-agnostic video segment-and-track where pretrained detectors have no labels, e.g. a neutrophil chasing a bacterium.
Method
Human-seed-then-propagate: a few clicks lock the right object, then SAM2 propagates masks across the rest of the frames autonomously. No training required.
Provenance
This is the sam2_hiera_tiny checkpoint from
facebook/sam2-hiera-tiny,
re-hosted here for the Model Garden showcase. All credit to Meta AI.
Usage
See the SAM2 repository for the video predictor API.
- Downloads last month
- 42
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
# Use SAM2 with images import torch from sam2.sam2_image_predictor import SAM2ImagePredictor predictor = SAM2ImagePredictor.from_pretrained(mfranzon/sam2-tiny-tracking) with torch.inference_mode(), torch.autocast("cuda", dtype=torch.bfloat16): predictor.set_image(<your_image>) masks, _, _ = predictor.predict(<input_prompts>)