ResNet

ECE-GY 7123 Deep Learning - Mini Project 1 - ResNet on CIFAR-10 within budget

PreRequisistes

The following python packages are required:

• torch
• torchsummary
• numpy
• tqdm
• multiprocessing

Install them manually or use this in your python notebook: ! pip install torch torchsummary numpy tqdm multiprocessing

Overview

• project_model1.py: This file is the core model that is used in this project.
• DataFetcher.py: This is used to load CIFAR-10 dataset and apply various Data Augmentation techniques.
• customTest.ipynb: This file is a sample test script developed initially for debugging purposes. A sample dataset of classes from the CIFAR-10 dataset was used defined by the urls array.
• test.ipynb: This file is used for testing on test dataset. It results in final accuracy of 93.16%.
• train.ipynb: This file is used for training the model on train dataset.

Results

Type	Accuracy
Training	99.9998%
Testing	93.16%