SAM Studio
Last updated on
I built SAM Studio to make state-of-the-art computer vision completely accessible, right in your browser. It’s a frictionless, privacy-first image segmentation tool powered by Meta’s Segment Anything Model (SAM 3).
Instead of sending your images to a remote server or setting up complex Python environments, to run the model: SAM Studio downloads the model directly to your browser’s local storage cache upon the first visit. From then on, all inference and masking logic runs 100% locally on your machine, ensuring your data never leaves your device.
How it operates
- Drop an image: Drag in any photo.
- Point and Click: The UI is hyper-minimalistic. Left-click on an object you want to extract to add a “positive” point. Right-click on areas you want to exclude to add “negative” points.
- Export: Instantly download the resulting binary Mask, or a transparent PNG Cutout of the isolated object.
Try it out
The application is deployed entirely client-side. Check it out at the link below: