Animal_Classification_DL

Animal Image Classification

This code trains a convolutional neural network (CNN) to classify images of 10 different animals.

Data

The data consists of images of 10 animal classes:

• Butterfly
• Cat
• Chicken
• Cow
• Dog
• Elephant
• Horse
• Sheep
• Spider
• Squirrel

The data is split into a training set and a test set. The image data is stored in the data/animals folder, with separate subfolders for each class.

Model

There are two model architectures implemented:

• SimpleCNN: A simple CNN with 2 convolutional layers and 2 fully connected layers
• AdvancedCNN: A more advanced CNN with 5 convolutional layers and 3 fully connected layers

The models are defined in models.py.

Training

The main training script is animal_train.py. It handles loading the data, initializing the model, defining the optimizer and loss function, training for multiple epochs, and evaluating on the test set.

Key parameters:

• --batch-size: Batch size for training
• --epochs: Number of epochs to train for
• --log_path: Path to save TensorBoard logs
• --save_path: Path to save trained model checkpoints

Use python animal_train.py --help to see all available arguments.

The script uses PyTorch and leverages GPU acceleration if available. Progress bars and TensorBoard logging are used to monitor training.

Evaluation

Model accuracy on the test set is evaluated at the end of each epoch. The best performing model checkpoint is saved.

Usage

To train a model:

python animal_train.py

This will train the AdvancedCNN model for 100 epochs and save checkpoints to trained_models/animal.

You can customize the model, hyperparameters, and output paths by modifying the commandline arguments.

Requirements

The code requires the following packages:

• PyTorch
• torchvision
• tensorboard
• sklearn
• tqdm

Use pip install -r requirements.txt to install the required packages.