Lucas comments

Results 8 comments of


                                            Lucas

Support Prompting with Vague Segmentation Masks for Refinement

@ZhengPeng7 Thank you for the thoughtful response! 🙌 I’m thrilled to hear that refining SAM-like masks aligns perfectly with your vision for BiRefNet’s future—it’s encouraging to see our ideas converge!...

The refine_foreground function takes a long time to execute; how can it be optimized?

@xiao-keeplearning You can try to use native torch functions instead of opencv functions, together with GPU acceleration, refine_foreground only costs maybe 100ms depends on your GPU devices. My code is...

The refine_foreground function takes a long time to execute; how can it be optimized?

@ZhengPeng7 Hi, I wanted to follow up on the previous code implementation. My apologies - I discovered a minor precision issue in certain edge cases. When using lower-precision inputs like...

The refine_foreground function takes a long time to execute; how can it be optimized?

@ZhengPeng7 No problem. maybe this weekends cause I have to work during the weekdays. I'll check your code

The refine_foreground function takes a long time to execute; how can it be optimized?

@ZhengPeng7 Sorry for the late reply. After testing your code, I did find some issues. I will first briefly list the problems and then explain them in detail. #### 1....

The refine_foreground function takes a long time to execute; how can it be optimized?

@ZhengPeng7 glad to help 😎

How you do zero-initialize and how good this technique is?

thanks for your replay! I thought at first you were doing zero gating like what LLAMA-Adaptor did.

How you do zero-initialize and how good this technique is?

by the way, the arxiv link of your paper in the homepage misleadingly directs to your another paper @hrz2000