BLIP
BLIP copied to clipboard
Confidence scores for image captions?
Hello - is there any way for the model to also return a [0,1] confidence score when generating an image caption?
Hi, you can pass in the generated caption to our ITM model to get a similarity matching score. You may want to refer to examples in our LAVIS library: https://github.com/salesforce/LAVIS/blob/main/examples/blip_image_text_matching.ipynb
Hello, What about the visual question answering? How can we get the probabilities? Thanks