image-classification-deeper-networks

Image classification using deeper networks like ResNet, VGG, GoogLeNet etc.,

Requirements

1. Run pip install -r requirements.txt from the project directory to install the requirements.
2. Data will be stored in wandb, so it should be installed and account has to be created. Also get the api key from wandb account online and add WANDB_API_KEY to the environment variables.

Run instructions

1. To train the model, run

   python main.py train -r <data_path> -d <dataset_name> -m <model_name> -s <model_save_path> 

   Models Implemented: resnet18, resnet34, resnet50, resnet101, resnet152, vgg11, vgg13, vgg16, vgg19, GoogLeNet
   Datasets Included: mnist, cifar10, cifar100
   Note: Download MNIST dataset manually and place it in the data_path as it's website is closed down.

2. To test the model on the test data, run

   python main.py test -r <data_path> -d <dataset_name> -m <model_name> -s <model_save_path> 

   Models Implemented: resnet18, resnet34, resnet50, resnet101, resnet152, vgg11, vgg13, vgg16, vgg19, GoogLeNet
   Datasets Included: mnist, cifar10, cifar100