Hire Xiangdong W. - Computer Vision Engineer

Design and customize deep learning algorithms for the company’s unique metasurface-based camera, including face detection, dense facial landmark detection, attention awareness, pose estimation, face recognition, spoof detection, etc. Design and customize computer vision and image processing methods for polarized images, such as near pixel-wise accurate pupil localization. Implement deep learning and computer vision algorithms in C++ on Linux/x86 and Android/ARM platforms. Quantize machine learning models trained with TensorFlow or Python so they can run on Qualcomm NPUs, e.g., Hexagon or GPU.

Designed and implemented computer vision functions for Polycom EagleEye camera, which might be the first camera that can automatically detect, track and zoom in person at any poses. Developed in-house computer vision algorithms such as human detection, lip reading and face recognition. Customized and optimized and the source code on resource-constrained platforms, increased the performance from demo to product mature level.

Developed and improved video codec algorithms to enhance DTV products.