Hi,
First thanks for your work on this project, it's really great!
In your paper, you mention a way to create segmentation masks explicitly controlled by user prompts, but this is not present in the code shared here, only the full auto-segmentation of a model is presented here. Could you share an example of explicit user prompt to generate a mask like this?
Is it also possible to merge the results of multiple prompts? The code mention labels per prompt, which can be 0 or 1 for a negative or positive point. Could you share an example of how to use this?
Thanks!
Hi,
First thanks for your work on this project, it's really great!
In your paper, you mention a way to create segmentation masks explicitly controlled by user prompts, but this is not present in the code shared here, only the full auto-segmentation of a model is presented here. Could you share an example of explicit user prompt to generate a mask like this?
Is it also possible to merge the results of multiple prompts? The code mention labels per prompt, which can be 0 or 1 for a negative or positive point. Could you share an example of how to use this?
Thanks!