Hi
I have used arcgis.learn on an aerial image, i.e. where you have a single image with many objects in it, on which you train a model and later use to detect similar objects in that same image.
What if you have hundreds of "ordinary" images i.e. non-georeferenced images taken by a camera from the ground? I have hundreds of photos of similar objects which I would like to classify and detect using arcgis.learn. However, when you use the "“Label Object for Deep Learning” tool in ArcGIS Pro, it is setup to draw around objects in a single (very large) image such as an aerial photograph. If you then export the labelled dataset it can be used by "Train Deep Learning Model” to train the model.
However there seems to be no facility to load, say 100 snapshots in ArcGIS Pro and create a single labelling dataset out of these 100 or so images for training purposes. Does ArcGIS Pro cater for this or is this not something the deep learning tools in ArcGIS Pro are supposed to be used for?
If this is possible, how would you do it? The question is not only about labelling multiple images and output that as a single labelled dataset, but after the model has been created, also detecting similar objects, not in a single image, but in multiple images.
Thanks
Hugo
