Coding Projects

Flower Recognition

An app for flower recognition

An easy iOS program, using Apple's Machine Learning Framework, snap a photo of a flower with the camera, and it will tell you what flower it is. The flower image classifier used was originally a Caffemodel and translated into Apple's Mlmodel using CoreML Tools on PyPI, the image you take will go throw the Mlmodel and return a result of string type, then the string is being used for an HTTP request to Wikipedia using Alamofire, the JSON result is parsed with SwiftyJSON, and the image is retrieved using SDWebImage.

Close Project

Ball Game

Ball rolling game created using Unity

I had considered the possibility of doing game development too. A game can be fun without 50GB of texture and particle effects, simple and engaging is enough.

Close Project

Snake Game

Our favorite snake game from Nokia!

This is one of the early projects I did when I first learned Python online, originally the code was laying around in the script but I changed it into the MVC model eventually.

Close Project

GED Project

A Python project from Python course

The final project for my Python class, a conclusion of all basic Python syntax and object-oriented programming, also simulated a real-life work environment where one works in a group with an Agile Release Train schedule and that everyone pushes their work into his or her branch.

Close Project

Dynamic Graph Network

Data visualization project for my summer internship

A data visualization project I did for my summer internship, using a front end model for data science called Plotly, one can select date and click on nodes to see relationships among them, the edge will show the weight between 2 nodes when hovered on, and a graph below to show their properties. (GitHub link is hidden)

Close Project

NYC Crime Prediction

A machine learning group project for an applied machine learning course

The final group project for an applied machine learning course, which predicts the crime trend for 5 Boroughs of New York City, crimes are divided into 3 major crime classifications: violation, misdemeanor, and felony. Crime data is sourced from the NYC Open Data website, we used 3 machine learning models for the training of data, consecutively they are Linear Regression, Neural Network, and Decision Tree.

Close Project

Weather Customization

An iOS weather app project from Human-Computer Interaction class

An iOS weather app that one can add or remove weather information tiles, or drag and drop to rearrange them. And when you click on the Voice button or lauch it with Siri, the app will read all the information of your choice to you.

Close Project

Pac-Man Path Search

A reinforcement learning project with beloved Pac-Man

(From top to bottom: Depth-First Search, Breadth-First Search, and Uniform-Cost Search)

An ongoing reinforcement learning project, the goal is to allow Pac-Man to teach himself to find the best path to eat all the dots while avoiding the ghosts using different search methods and reinforcement learning.

Close Project

Stevens Repository

A Flask, Jinja, SQLite project from Python class

A database project for the Python class, I used SQLite as a local, file-based relational database management system, and Flask as the web server, to record Stevens's personnel.

Close Project

Space Exploration

An OpenCV computer vision Python game that reads hand gestures

It's a temporary backup plan I came up with for a capstone project, and it utilizes OpenCV's contour detection functionality to decide whether it is a palm or a fist on both sides of the screen. The game will rotate left if I clench my left hand and rotate right if I clench my right hand, and will move forward if both hands are clenched, I made the prototype in 1 day so the codebase isn't at its fullest glory.

Close Project

Poke3D

An iOS AR demo that puts pokemon on their trading cards

A small project to get used to AR concepts like 3D scenes, modeling, image tracking et cetera.

Close Project

Athena AI

Chat with ancient philosophers or anyone you wish.

Imagine you can talk to anyone you want, such as philosophers, or fictional characters, the project takes a vocal input, transcribe to text, then use Lancaster stemming algorithm to turn a paragraph into a list of stems, then a trained DNN is used to determine whether the question the user asked belong to a prewritten topic or a deeper question that can only be answered with GPT 2 Natural Language Generation (Thanks Elon).

Close Project