Deep Learning Image Classification challenge on custom image dataset which contains nearly around 3700 images as training set and around 170 images for validation. The task of Image Classificatio is solved by creating and training a Convolutional neural Network which contains CNN Layer, Skip connections, BatchNorm, MaxPool, Dropout and a Fully Connected Linear Layer . With this network architecture, I achieved an accuracy of 88%

Framework used: PyTorch

Hyperparameters: Batch size = 32, Number of Epochs = 25, Learning rate = 0.001

Loss Function: CrossEntropyLoss

Optimizer: ADAM