This project is a GUI (Graphical User Interface) for classifying dog breed by images. It uses a pre-trained neural network called ResNet-18 to perform the classification.
In the develop mode, the code has 3 main functions:
1. `download_images()`: Download images of dogs from the internet and store them locally, using AZURE BING SEARCH API.
2. `train_model()`: Trains the dog image classification model using transfer learning on top of the pre-trained ResNet-18.
3. `classify_image()`: Uses the trained model to classify a dog image.
In the default mode, the code has only classify_image(), and the model is downloaled automatic.
The code also has a Gradio interface that allows the user to load a dog image and get the model's classification. The model classifies images into 13 different dog breeds:
The interface also has a memory optimization option that can be enabled or disabled by the user. When the option is enabled, the template uses less memory but takes a little longer to sort the image. The option is disabled by default. User
