shersoni610

Hello, I changed "cuda" to "cpu" in all the gradio files. Although nothing crashed, the code runs forever without producing results. Is is impractical to use MAC M1 cpu? Somehow,...

Error in running openset example

1

Hello, I am trying to run inference with openset. I get the following error: checkpoint, the shape in current model is torch.Size([768, 512]). size mismatch for tagging_head.encoder.layer.1.crossattention.self.value.weight: copying a param...

Hello, in the "inference_ram_openset.py" looks like the default mode is tag2text. Should it be ram model? Thanks

Image sizes

4

Hello, what image sizes are supported by ram and tag2text models?

Inference_ram.py example using Tag2Text

1

Hello, I am confused. It seems ithe nference_ram.py example is using default='pretrained/tag2text_swin_14m.pth' which is the same as the model used in the inferecen_tag2text.py. Can someone clarify if this was not...

Hello, How do I get results of auto mode using points or boxes? Assigning points on grids as displayed on SAM website gives only one mask for the entire image....

Hello, I have an image containing many fruits. When I use Auto mode then I get pretty good results and I get different mask for different a fruit. However, when...

Hello, On Mac m1/m2 there are issues with tkagg backend, so I was wondering if there is a code which uses only opencv for plotting and avoids matplotlib lib. Thanks

Hello, I see that viewing annotation has been defined in both the prompt.py and utils directory. Which one should we use? On MAC M1, I do get segmentation results with...

Apple MPS support?

1

Hello, Does anyone knows how to use apple "MPS" instead of cpu/cuda device?