I am working on an image classification problem, the input data normally is images to classify, but I thought latitude and longitude would play something on these satellite images. I sorted by latitude, but I could sort by x and y coordinates and applied an LSTM model, that proved coordinates has an influence on target.

How to add this as a neural network (LSTM) to the more important image classification neural network. Any hint or suggestion would be much appreciated.

  • 1
    $\begingroup$ I didn't figure out your question completely but you may want something like this. $\endgroup$ Commented Dec 24, 2018 at 19:16
  • $\begingroup$ Check kaggle competitions for more ideas :) if I got you questions l correctly, you must train your CNN to output two things then... One the probability and secondly the latitude and the longitude it thinks the image is from maybe $\endgroup$
    – Aditya
    Commented Dec 24, 2018 at 20:06
  • $\begingroup$ @Aditya you didn't get my question, latitude and longitude are features among image data, orientation is the target. $\endgroup$
    – bacloud14
    Commented Dec 25, 2018 at 2:24
  • $\begingroup$ Yep! I forwarded your question to get help! Will update if someone shares anything that can help you! $\endgroup$
    – Aditya
    Commented Dec 25, 2018 at 2:31
  • $\begingroup$ @Aditya Media's suggestion seems the right one, I will be back with something . ty $\endgroup$
    – bacloud14
    Commented Dec 25, 2018 at 2:34


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.