Machine Learning Engineer
Current- Research, design, implement and maintain the safety violation reporting system using YOLOv5 object detector and FastMOT multi-object tracking framework- Design and implement multiple detectors that report potential hazardous behaviors in working environment- Design a method to create 3D cuboid data from 2D images for model training and proceed to predict 3D cuboids from 2D objects- Implement and optimize performance of processing videos with near real-time performance based on Docker and GPU technology- Analyze and implement system to retrieve real-world data by using only surveillance camera videos- Using AWS cloud services to develop, transform academic deep learning methods to production software