This repository contains all of my machine learning projects from Summer 2021. A brief description of each project can be found below. I tried to explain what I as doing in detail throughout each project.
An ASL Classifier neural network that can detect signs from the American Sign Language alphabet, including space, del, and nothing. It uses a standard neural network architecture and achieved 97% accuracy in mmy tests. Potential future work includes real time integration with Snapchat via Lens Studio
Deep Convolutional Generative Adverserial Network (DGCAN) designed to learn to replicate landscape paintings from this dataset.
Architecture based on this tutorial, but many modifications were made along the way.
Trained for 6 hours, totalling 36 epochs in Google Colab. Runtime kept disconnecting, so I could not get super great results. I think it could do better with more training, but I want to try reworking it as a Progressive Growth GAN (PGGAN), found in another Colab file.
The same model as the Colab notebook, but I converted the code to traditional python stuff. I ran it a couple times for 2.5 hours, totalling 300 epochs and achieved decent results actually.
Final results from the best round of testing
More images can be found in various version folders within DCGAN.
Progressive Growth Generative Adverserial Network (PGCAN) designed to learn to replicate landscape paintings from this dataset.
Architecture based on this tutorial. Still a work in progress. Once I get it working, I'm hoping to make some changes to it to make it mine
Standard python version of PGGAN.ipynb. Improved architecture, statistic collection, etc. (WIP)
Simple convolutional network built in Pytorch to identify and classify images from the CIFAR10 dataset
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent