Course Projects

This survey explores recent advancements that address the challenge of uncertainty quantification and dependence on explicit priors in black hole image reconstruction. We focus on methods that reconstruct black hole images with uncertainty without relying on explicit priors. We'll explore how these methods leverage generative models within a variational Bayesian framework to overcome the limitations of traditional approaches. We'll also delve into the limitations of these recent methods and propose promising avenues for future research.

This is a non-deep learning fire detection pipeline inspired by this paper. Our pipeline comprises of three parts: color space classifier, color component classifier and texture classifier. Our model was trained and tested on the BoWFire Dataset and is able to detect fire from static images with an accuracy of 80%.

The Transiting Exoplanet Survey Satellite (TESS) is an all-sky survey mission led by MIT and NASA in search of exoplanets. A certain number of planet candidates (PC) discovered by TESS are not actual transiting exoplanets, which are usually referred as "false positives", therefore follow-up ground-based photometry is performed for verifications. We conduct ground-based transit photometry on two planets candidates declared in TESS, TOI-5278.01 and TOI-4439.01 with an 80cm telescope at Xinglong Observatory, P.R. China. Our data processing pipeline comprises of three parts: image reduction, aperture photometry and EXOFAST fit. For TOI-4439.01, our result is in accordance with TESS data, proving the exoplanet nature, while the EXOFAST fit did not converge for TOI-5278.01 due to low SNR of the light curve.

This repo contains my code and data for astronomical image reduction. The images were taken at Huairou, Beijing with a 25cm telescope of DoA. Targets include M13, M51, M3, etc.

Developed three algorithms that solve the following problems respectively:
1. Recognize faces from videos clips.
2. Recongizes voices from audios.
3. Given a video of three speakers speaking at the same time, separate the speeches from the three speakers.
This project is implemented in Python with the aid of Face Recognition, Resemblyzer and Speechbrain.

Built an algorithm using Guassian Process Regression to predict the stock prices from previous observations. This project is implemented with Python and Scikit Learn.

Developed an algorithm to create a video from a given set of video clips and a piece of background music to best match the clips' audio rhythm with the background music. This project is implemented with Matlab.

Website template from Dr. Jon Barron