Segmentation with the ml backend sometimes very slow over the network
The network connection must be very stable and fast so that the segementation works smoothly (the images then also take longer to load). As soon as the internet slows down (often with WLAN), the data transfer takes a very long time, which leads to considerably delayed segementation in the frontend.
Suggestion: Since the images are pulled via the IIIF image API, the user could have the possibility to manually adjust the size of the images in the website and, for example, reduce them if the network connection is not so good. Then the images will also load faster.