This repository consists my simple solution for the kaggle https://www.kaggle.com/c/dogs-vs-cats-redux-kernels-edition challenge
This solution scored a logloss of 0.06499 on the public leaderboard
-> Several keras pretrained models like Resnet-50, Resnet-102, InceptionV3 are used with finetunning.
-> Ensembling different models increase model accuracy significantly.
-> Aggresive Dropout on Dense layers has been used to reduce overfitting.
-> Images are being pre-processed to and reduced to size (224, 224, 3).
-> diffenrent data augmentation like sheering, zooming, horizontal-vertical shifts and flips are done.
-> PReLU activation gives best result