This repo is a packaged version of the segment-anything model.
pip install metaseg
from metaseg import SegAutoMaskGenerator
# If gpu memory is not enough, reduce the points_per_side and points_per_batch.
# For image
autoseg_image = SegAutoMaskGenerator().save_image(
source="image.jpg",
model_type="vit_l",
points_per_side=16,
points_per_batch=64,
min_area=0,
)
# For video
autoseg_video = SegAutoMaskGenerator().save_video(
source="video.mp4",
model_type="vit_l",
points_per_side=16,
points_per_batch=64,
min_area=1000,
)
- Support for video files
- Support for pip installation
- Support for web application
- Support for automatic download model weights