Dog vs Cat Convolutional Neural Network classification

Data

The model is trained on Kaggle dataset kernels edition

Model

We use Tensorflow and Keras MobileNetV2 to build the model. The last layer is sigmoid with 2 classes (dog and cat). See jupyter notebook for more details.

Training

• The data from the Kaggle dataset is processed to fit into MobileNetV2 input. See jupyter notebook for more details.

• The model achieved 99% accuracy on train set and 95% on validation set after 5 epochs