pet_image_classification_system

Pet image classification system written in Python with PyTorch.

Source of the Oxford-IIIT Pet Dataset

• The Oxford-IIIT Pet Dataset
• OxfordIIITPet - PyTorch

Usage Instructions

1. Train a Model

1.1 Start TensorBoard to provide the visualization required to check how well the trained model did:

tensorboard --logdir=runs

1.2 Train a model:

python3 train_model.py

2. Pick a Pet Image

2.1 Use the GUI program to pick a pet image:

python3 gui_pick_pet_image.py

3. Run Program

3.1 Run the program with the file path of the saved model:

python3 main.py trained_models/model_20241110_144454_22.pt

• The file name format of the saved models:
  • model\<timestamp>\<epoch_number>.pt"
• The timestamp format:
  • %Y%m%d_%H%M%S

Note: It is a good idea to pick a model with the highest number of epochs.

Resource

• PyTorch Website
• Introduction to PyTorch
  • Learn the Basics
  • Quickstart
  • Tensors
  • Datasets & DataLoaders
  • Build the Neural Network
  • Optimizing Model Parameters
• Introduction to PyTorch on YouTube
  • Introduction to PyTorch - YouTube Series
  • Introduction to PyTorch
  • Introduction to PyTorch Tensors
  • The Fundamentals of Autograd
  • Building Models with PyTorch
  • PyTorch TensorBoard Support
  • Training with PyTorch
  • Model Understanding with Captum
• Nesterov Momentum Explained with examples in TensorFlow and PyTorch - Medium
• Pytorch Change the learning rate based on number of epochs - StackOverflow
• How to split data into 3 sets (train, validation and test)? - StackOverflow
• Oxford IIIT Pets Segmentation using PyTorch - GitHub
• RuntimeError: stack expects each tensor to be equal size, but got [3, 224, 224] at entry 0 and [3, 224, 336] at entry 3 - PyTorch Discuss
• Make a Simple Image Gallery Using Tkinter and Python - Medium
• Loss Functions in PyTorch Models - Machine Learning Mastery
• timm/oxford-iiit-pet - Hugging Face
• IndexError: Target 2 is out of bounds
• How do I use a saved model in Pytorch to predict the label of a never before seen image? - StackOverflow