In this project, we implemented both neural network modle and physical reflection model to esimate the occupancy of a space for smart lighting. We use perturbation-modulated method to compute the light transport matrix, through which we can estimate the occupancy.
My work is: (1) Transfer the system(reflection model) from testbed to our much bigger smart conference room. (2) Develop the algorithm for the regression method with neural network. (3) Implement a new multidirectional "compoud eye" sensor.
The website of our project.
OCCUPANCY DETECTION BASED ON COLOR SENSORS
Computer Vision/ Machine Learning/ Smart Lighting/ Occupancy Estimation/ Color Sensor
Rensselaer Polytechnic Institute, Troy, NY
This is my final project for Prof. Arthur Sanderson's course Pattern Recognition. In this project, I showed a demo based on neural network to classify the genre of a song, after listening the song for 30 secones. Demo
MUSIC GENRES RECOGNITION
Machine Learning/ Patter Recognition/ MATLAB
Rensselaer Polytechnic Institute, Troy, NY
This is my final project for Prof. Qiang Ji's course Computer Vision. I set up a system to recover the 3D properties of the object using a laser line based on the active stereo vision technique. I implemnted a method to compute the plane of the line laser, so the laser is free to scan over the object from multiple angles. Download the report here
3D SCANNER
Computer Vision/ MATLAB
Rensselaer Polytechnic Institute, Troy, NY
This is my master's project in Boston University.
PLL and 3D SUBSTRATE NOISE MODEL
Analog/ RF/ Cadence
Boston University, Boston, MA