Yunqian (Patrick) Cheng has been working with Micro-AV’s project team since fall 2020. Their contributions have led to major progress in both hardware and software for the Millipod systems. Developing training algorithms and improvements for vision and signal-processing systems in object detection using YOLO and other software. Patrick has helped implement mathematical models used in lane detection as well as lead team research into 2D LIDAR data collection and processing.
We here at Micro-AV would like to thank Patrick for their hard work and efforts; here’s a word from our featured participant themselves:
“Hi, my name is Patrick Cheng, an Informatics undergrad from UW who is graduating in Spring, 2022. My research focuses on computer vision and graphics as well as their application in accessible technologies, including autonomous vehicles, mobile health care, and augmented/mixed reality. I have been working with Dr. Folsom as a computer vision researcher at Micro-AV since September 2020. I recently joined the multi-task temporal shift convolutional attention network (MTTS-CAN) research team with Xin Liu from UW and Daniel McDuff from Microsoft as well. I am looking forward to getting my PhD degree in computer science and working as a computer vision research developer in the industry.”
Key Contributions
- Trained and optimized YOLOv5-s object detection network.
- Researched ways to improve YOLOv5 robustness with small-dataset training.
- Performed image segmentation and lane detection with SegNet, Canny Edge Detection, and Hough Transform.
- Performed Bayesian 2D Lidar signal processing to implement autonomous vehicle positioning and obstacle detection from a laser scan
- Explored the latest computer vision papers and information, collected the key training data.