This is a github repository is a record for the work me and my partner(Yibo Kong) have done in the CS639 in University of Wisconsin in Madison, in year 2024, which is well adapted from Stanford’s CS231n (CNNs for Visual Recognition) and Michigan’s EECS 498-007/598-005 (Deep Learning for Computer Vision). These projects involve implementing classical and fundamental deep learning algorithms/architectures/strategies (including neural network forward and backward pass) from scratch.
Pytorch Tutorial
Image Classification, k-Nearest Neighbors (kNN), Support vector machine (SVM), Softmax, ReLU, Fully-Connected Neural Network
Fully-Connected Networks, Convolutional Networks, Batch Normalization, Kaiming Initialization, Max Pooling
Feature Pyramid Network (FPN), Fully Convolutional One-Stage Object Detection (FCOS), Box/Centerness Regression, Non-Maximum Suppression (NMS)
Generative Adversarial Networks (GAN), GAN Loss, Deeply Convolutional GANs, Style Transfer, Total-variation regularization, Gram Matrix