Efficient Track Anything (EfficientTAM)
Interactive Video Segmentation
Built our demo based on
SAM2-Video-Predictor
. Thanks to Sylvain Filoni.
Instruction
Upload one video or click one example video
Click 'include' point type, select the object to segment and track
Click 'exclude' point type (optional), select the area you want to avoid segmenting and tracking
Click the 'Segment' button, obtain the mask of the first frame
Click the 'coarse' level and the 'Track' button, segment and track the object every 15 frames
Click the corresponding frame to add points on the object for mask refining (optional)
Click the 'fine' level and the 'Track' button, obtain masklet and masked video
Click the 'Reset' button to restart
Github
link
Input Video
▼
Input Video
Drop Video Here
- or -
Click to Upload
point type
include
exclude
Clear Points
input image
Frame with Point Prompt
Checkpoint
Segment
Try some of the examples below ⬇️
Examples
Frame ID
change current
Reference Mask
Track level
coarse
fine
Track
Reset
Masklets
Video