Segment Anything (Meta)

Last update time : 2025-09-04 22:34:53

Segment Anything (SAM) is an advanced AI model that can cut out any object in any image. It features zero-shot generalization, allowing it to perform successfully on unfamiliar images without requiring prior training.

Developed by Meta, the Segment Anything Model (SAM) is a robust artificial intelligence system designed to segment any object within an image. One of SAM's most notable features is its ability to use various prompts, such as a simple click or a bounding box, for flexible integration with other AI systems. This adaptability enables it to work as a general-purpose segmentation tool rather than being limited to a single task. SAM excels at zero-shot generalization, performing effectively on images it has never seen before. The model is also used to automatically annotate large-scale datasets and continuously improve them. The project's core contributors include Alexander Kirillov, Eric Mintun, and Nikhila Ravi. The code is available as an open-source project on GitHub.

Pricing : Open Source

Web Address : Segment Anything (Meta)

Tags : AI segmentation object recognition image processing open source photo editing