Added image annotation based on your code, and combined training, prediction, etc. with the addition of a UI.

[sam-hsueh]

https://github.com/sam-hsueh/ImageSegmentation

[IntptrMax]

https://github.com/sam-hsueh/ImageSegmentation

Good job!

[sam-hsueh]

https://github.com/sam-hsueh/ImageSegmentation2

[sam-hsueh]

With Your SAM Code..

[IntptrMax]

https://github.com/sam-hsueh/ImageSegmentation2

MobileSAM maybe better than vit-b in this scenario. It can be faster and use less vram, and get a very similar result. Please have a try.

[sam-hsueh]

Thank you. I'll give it a try.