How to detect roof top objects in images using pytorch

I am little new to pytorch. I have see many example of object detection using tensforflow , Keras but could not find one which can give me some pointers for object detection in roof top images. any pointers , advise.

I assume you are looking for object detecting and localising (bounding box). If so, you can have a look at this example from pytorch here, . However, this is a fine running example and consists of block diagram of how object detection and localisation works.

If not try to explain the scenario a bit more contextually so that community may suggest one

1 Like

@nivesh_gadipudi , I am looking to train the model using my custom data set which has trees, houses and some other objects. I want to identify mainly trees from those images. the example you gave does not use the custom data set. I had gone through it before. Thank you for your response.

For a custom dataset, you can inherit class with
len(returns length of your dataset), getitem(takes in the index) and you would need to use some files libraries and image ones to open and load files from your custom dataset.

You can check these links out:


Hopefully, these links will be helpful.

1 Like