Hello everyone, I have a dataset of Arabic handwritten text images, and our dataset is labeled using text files describing every written letter in every line. we use this ground truth text file to make sure our bounding box labeling of the images are correct. But, we need to know the image name in order to lookup it’s ground truth. I have looked in the docs for something that might help, but no luck.
We use the asset metadata to do this. You can provide this as a text string to show, or a link to an asset which can even be HTML is my read on it. We have done it with text so far and it is very powerful just to have the info right at hand. In the labelling app there will be a little button in the upper left when this is available for an image.