This project aims to get more familiar with Keras and TensorFlow libraries to build a neural network model for predicting the type of animal in a given image using the Oregon Wildlife dataset, including four categories, namely elk, raven, bald_eagle, and raccoon.
- Preprocessing, including grayscaling the images and one-hot encoding of labels
- Neural network implementation
- Data classification and evaluation of the impacts of optimizer, number of epochs, loss function, and regularization
- Dimensionality reduction for data visualization