I'm an engineering studying at New York University. I am interested in robotics, computer vision and software development.
🔗 GitHub
(Aug 2022 - Present)
Developing low power pedestrian counting in AI4CE lab for NYC Office of Technology&Innovation
– Deployed on Coral Dev Micro. Using TensorFlow Lite (Deep Learning) model for pedestrian detection.
– Using Gaussian Process Regression (Machine Learning) model to predict pedestrian count during off duty cycle.
– Developing AWS back end and dashboard.
(Jan 2022 - May 2022)
Based on pervious published work, This project repeats the 3DMM paper experiment. We trained a neural network to get the αid, αexp ,αalb ,γk ,T , R parameters from the input image. Before getting all the coefficients from that Deep Neural Network, we extracted the landmarks from input dataset by a face alignment detector for later loss calculation. After obtained the coefficients, we used BFM to get all the matrices needed for 3D face rendering. Soft-Rasterizer used these matrices to compute the mesh and render the mesh as a 2d image. We optimized the model by reducing the loss based on the output image and mesh.