Detecting and Classifying waterbirds in aerial imagery with Deep Learning tool

Question

I am a PhD student and I am trying to train a deep learning model to detect and classify waterbirds in aerial imagery using the tools in ArcGIS Pro. I have tried the pre-trained model for tern detections in the living atlas and it is not working well on the imagery I am using. I have over 300 images of each species I am to identify and classify already bound in the labelling process of GIS. I have tried exported these as Pascal Visual Object Classes and running it with a Faster R-CNN model in GIS but the model will not run. I am using imagery of 96dpi resolution and image size exports at 256 X 256 pixels but am unsure if the rest of my setting are appropriate for what I am trying to do?

Any help would be greatly appreciated!

UmairWaris257 · Answer

This is a very interesting use of deep learning in real life. It helps us find and identify waterbirds in aerial images easily. This can also help protect nature and study birds better.

Marissa226 · Answer

Thanks for the quick response @Kate_Fickas ! I think that connection would be great! We already have some work we have done that we would love to incorporate into a GIS dlpk so any help would be greatly appreciated!

Kate_Fickas · Answer

Wow, what a cool application @Marissa226! We have a team member who is working on classification of individual animal species. Let me wrangle him into this conversation! Stay tuned...