This project is an implementation of a machine learning model that predicts the breed of a dog given an image of the dog. The model was built using a dataset of dog images labeled with their corresponding breed, and was trained using convolutional neural networks (CNNs).

The dataset used for training and testing the model is the Stanford Dogs Dataset, which consists of 20,580 images of 120 different dog breeds. The images are split into a training set of 12,000 images and a validation set of 8,580 images.

The model used for predicting dog breeds is a CNN with multiple convolutional and pooling layers, followed by fully connected layers and a softmax activation function. The model was trained using the Adam optimizer with a categorical cross-entropy loss function.

The model achieved an accuracy of 96% on the validation set, which is a good performance for a multi-class classification problem with 120 classes. However, the model can still be improved by using more advanced architectures, such as transfer learning or ensembling.

#Usage To use the model for predicting the breed of a dog given an image, simply run the predict.py script and provide the path to the image file as a command line argument. The script will load the trained model from disk and output the predicted breed and its probability.

Python 3.x Pytorch NumPy Matplotlib