This is my finished project for my course Introduction to Machine Learning taught by Prof. Wang Jie at USTC. It is an simple implementation of ResNet34 using Python Numpy. It supports the training and inference process on part of tiny ImageNet dataset (in the folder dataset). If you need more information about dataset or the course please refer to http://staff.ustc.edu.cn/~jwangx/classes/210709/project.html
Run the script train.py to train on offered dataset. You can set hyper parameters in the script train.py. Also, if you would like to train on your own dataset, you must prepare a .txt file in the format of train.txt or test.txt first to enable the program to be access to your data. Notice this project does not support GPU computing so the training process can be very long on big dataset.
We have offered trained model in the folder model. You can run the script demo_test.py to test it on our test set.