Hello I am trying to cut a raster image into smaller chips with a labelled vector points to feed into a CNN network for regression. Can FME help in this regard?

Hello @odebs​ , thanks for posting to the FME Community! Yes, this should be possible in FME. A Clipper transformer can be used to divide your original image into smaller sections. From here, assuming no other raster processing needs to be done, a VectorOnRasterOverlayer will help get any vector data onto an existing raster background (eg. aerial imagery or orthophotos). As for regression, you could always explore the RCaller or PythonCaller (if you have a preference), but I am sure there are other potential methods too. We don't yet have a regression analysis transformer, but an idea has been made already! If this is something that could be useful to you, I recommend upvoting! Hope this was helpful, if you have any other questions - let me know! Best, Kailin