Trasnsfer_learning_dog_breed_classification
This program describes the use of transfer learning to build a model which can classify dogs with their breeds. It also compares the model which is trained from scratch, with the model which uses transfer learning. I observed test accuracy for test data to be around 2% while training model from scratch using 5 epochs, also it took around one hour (i am using a cpu), to train for the 5 epcohs, on the contrary side, whlile using model with transfer learninig ( named vgg_model). it reached an accuracy of 75 on test data, and the best part is that it took less than 20 second to run 20 epochs. This is apparently my first model that i have built using transfer learning. I reffered to this github file - https://github.com/denizn/Dog-Breed-Classifier/blob/master/dog_app.ipynb for this program. :)