The model is trained on Kaggle dataset kernels edition
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.
-
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