Animal-Classifier-with-ResNet

PyTorch implementation of a neural network to classify images of animals using ResNet. The code was written for the VLG Pixel Play Challenge and utitlises datasets provided in the contest for training.

Features

1. ResNet18: the model is trained using the ResNet18 architecture, using its pretrained weights.
2. Hyperparameters: The model is currently set to train for 15 epochs, which was obtained after some tuning.

Setup

1. Clone the repository.
2. Ensure you have all the Python packages installed. If not, run pip install -r requirements.txt in your virtual environment.
3. Download the dataset.
4. (or) You may run this code on Kaggle itself.

Run

Firstly, you will have to modify train_dir and test_dir in your code depending on its location.
Then run the code using python3 resnet.py