Defect inspection vision system for factories: Develop a user-friendly, deep learning-based vision tool for factory environments using either PC-based or embedded systems. By leveraging the latest models and domain knowledge, it can achieve high accuracy with relatively less training data and a low-latency inference server.

2-in-1 mask detection device: Integrated with Microsoft Azure and NVIDIA Jetson to enable real-time body temperature and mask detection. Completed in two weeks.
English news: https://ppt.cc/fUDiMx
Chinese news: https://news.microsoft.com/zh-tw/cth-ms-ai-mask-temparture-detector/

Population-Based Augmentation: Contributed a visualization example to check parameter results after training.
https://github.com/arcelien/pba

Electric car power managements circuit board test: Discuss with client about test plan. Select the measuring instrument for customer. Using programmable power supply, DMM and oscilloscope to build an automation test software for the customer via LabVIEW.

Picoamp test: For a customer needing to measure 200 pA , I selected a low-noise amplifier and a DMM to achieve 1 pA accuracy. Measurements were taken in a semiconductor factory for a MEMS component used in cars, via a GPIB connection.

Wi-Fi modem test: Utilized programmable power supplies, function generators, DMMs, oscilloscopes, and RF testers to evaluate modem functions. Tests included current, voltage, multi-frequency, and Wi-Fi signal functionality.

1 Power : Created a side project that allows you to monitor your power bank’s battery life via your phone. Used ESP-32’s analog input to measure voltage changes and displayed the data on a web server.
Projects and demo video: https://github.com/prob1995/1power