Open chenyuanjiao342 opened 1 month ago
Can you show examples of the input image and coordinates you're using and the predictions you get?
Can you show examples of the input image and coordinates you're using and the predictions you get?
The first image is the input image and scribbles, and the second image is the predicted result.
hmm I am not quite sure what's going on. Here are some ideas for troubleshooting:
Could xy be switched between your scribble mask and your image? I see you're swapping the last 2 dimensions of your input image. Have you tried visualizing the scribble mask and input image right before inference to check they are aligned?
Have you tried drawing a similar scribble on your image in our huggingface demo? You can clone the code for the demo here if you prefer to run locally. If you're able to get better results in the gradio demo, it suggests there's a bug in your code.
Hi, when I try to instantiate ScribblePrompt-UNet and make predictions, the segmentation results I get are not accurate. Part of my code is as follows. Is there anything wrong? Thank you very much!