I want to make a system that recognizes a single object using keras. In my case I will be detecting car wheels. How do I train my system just for 1 object? I did classification task before using cats and dogs, but now its a completely different task. Do I still "classify", with class 0= wheels, class = 1 non wheels(just random images of anything)? How do I do following steps in this problem? 1) Train system for 1 object 2) Detect object(sliding window or heatmap)
Your task is a so-called binary classification. Make sure, that your final layer has got only one neuron (e.g. for Sequential model model.add(Dense(1, ... other parameters ... ))) and use the binary_crossentropy as loss function.
Hope this helps.