Help with image recognition

Hello! I am a new programmer that wanted to get into machine learning. Recently, for my employment my boss asked me to modify the classification interactive to be able to upload photos to it instead of just taking pictures with the camera. I tried several options, but none of them worked. I believe that its because when I look at the camera output vs. picture, the camera is surrounded with several brackets to format, while the image is just pixel by pixel.

If anyone has any help slightly modifying the example they give you to accept uploaded photos, that would be great.

If not, and I have to do it from scratch thats OK too. I just would like some guidance on what needs to be done.

Thank you.

Hi @bglaenger, if you use PyTorch to train the classifier from https://github.com/dusty-nv/jetson-inference/blob/master/docs/pytorch-transfer-learning.md, you can put your images in a directory format like so:

‣ train/
	• class-A/
	• class-B/
	• ...
‣ val/
	• class-A/
	• class-B/
	• ...
‣ test/
	• class-A/
	• class-B/
	• ...

where class-A, class-B, ect. are the names of your image classes (e.g. cat, dog, ect.)

See here for more info about the directory structure of classification dataset: https://github.com/dusty-nv/jetson-inference/blob/master/docs/pytorch-collect.md