https://rohitd3.github.io/manyFacesML/
Our project is a surface level of Pytorch demonstration, and we intend to use it for our own learning purposes as well as yours. As you look through our code, we will try to supplement each segment with supporting blogs to explain as much as we can.
Tutorials and Guides Watched ~ 4 hours each
- https://www.kaggle.com/code/salmaneunus/image-classification-in-pytorch-cifar10
- https://www.kaggle.com/datasets/veeralakrishna/butterfly-dataset
- https://www.kaggle.com/datasets/arnaud58/flickrfaceshq-dataset-ffhq
- https://www.kaggle.com/datasets/andrewmvd/animal-faces
- https://github.com/jayrodge/Binary-Image-Classifier-PyTorch
- https://github.com/jayrodge/Binary-Image-Classifier-PyTorch/blob/master/Binary_face_classifier.ipynb
- https://hackernoon.com/binary-face-classifier-using-pytorch-2d835ccb7816
- https://github.com/jayrodge/Binary-Image-Classifier-PyTorch/blob/master/Binary_face_classifier.ipynb
- https://stackoverflow.com/questions/52468956/how-do-i-visualize-a-net-in-pytorch
- https://www.youtube.com/playlist?list=PLqnslRFeH2UrcDBWF5mfPGpqQDSta6VK4 ~ 8 videos watched
- Installation
- Tensor Basics
- Model, Loss, Optimizer
- Linear Regression
- Dataset and Dataloader - Batch
- Dataset Transforms
- Convulutional Neural Network(CNN)
- Saving and Loading Models
| Names | Hours | Commits |
|---|---|---|
| Everitt | 19 | Commits - Added blogposts explaining what PyTorch is, explained in between code, wrote code for displaying imagery and importing assets, dataset management, scrum management, markdown management, established dataset directory in google colab, installation of PyTorch, and bug/error handling in team code and own system. |
| Rohit | 16 | Commits - Found helpful links (blogs/videos) that group could use to research. Utilized kaggle to find large datasets that could be used to train/validify/test CNN model. Explained the use of transformations to Samuel and Nathan. Made graph that details overfitting or underfitting in the model and also explained what could be addressed to fix those. Added calulating loss in the training and valid data. Graphed epoch results. |
| Nathan | 17 | Commits - Worked with showing images and the transforms of the images. Researched and used plt or matplotlib.pyplot in order to show a visual demonstration of the transforms and data that we used. Worked on applying rotations and blurs to data and seperated input images into testing and training data. |
| Jun | 19 | Commits - Worked on test accuracy and improving the amount of loss for training. Figured out necessary imports (optim module with the use of CrossEntropyLoss()) for optimization algorithms, initialized the optimizer with a learning rate. Also put together Jupyter Notebook to explain my work. |
| Samuel | 16 | Commits - Helped Rohit make the graphs for loss data, explained the graphs, code for training neural network, a bunch of troubleshooting and error fixing for usingpytorch notebook file (imagesize, batch size, range, training, label, class size), explained training neural network code |