We are developing several cnn models like in tensorflow and pytorch but we are facing same problem in both cases... It's like we made our model train them even we got accuracy around 96-97. We also test few images before saving the model and that worked fine too. But after saving that model, we are testing it through Python code by which giving the saved model and path of a testing image from our data, it is showing same output, no matter how many images we changed. It happened with tensorflow so we shifted to pytorch but we are facing the same issue and as much as we know our testing code is also right so how to solve it? Continue reading...
