I’d like to be able to generate an image from data stored in metadata.csv when the corresponding point is selected / on-click in the tensorboard embedding projector.
Above, we can see that when a point is selected the word and count of occurences are shown in a small window. In my use case, I would like to have this window also display an image generated from the data using a custom function.
The motivation for this is to reduce necessary storage resources of storing an image for every single point when most will not be of interest. It appears to me that this is not currently supported, but if it is, please refer me to the relevant source or documentation. Otherwise, any advice on steps towards implementating this would be much appreciated. Thanks in advance!