AIoT Lab Projects
AI Wings - Lightweight AIoT Drones
AI Wings is an AIoT Drone System, which enables fast prototyping and test of 4G/5G AI drones. We create a embedded system module to allow users to convert general small drones into AIoT drones easily. To test AI Wings functions, we built a 80x80 cm delivery drone. (project home)
Drone Action Recognition
In this project, we develop a large-scale action detector of drone that can recognize numerous complex human actions such as running, eating, smoking, photo shooting, and create a large-scale dataset to evaluate our algorithm. (project home)
VIVID - Virtual Environment for Visual Deep Learning
VIVID is a photo-realistic simulator that aims to facilitate deep learning for computer vision, which supports different characters: robot, drone, and automobile. The platform can be used for many research fields including deep reinforcement learning, semantic segmentation object recognition and human action recognition. For more details please visit the project's homepage.
Pothole Detection
We work together with Feng Chia University to develop a real-time pothole detector on Xilinx FPGA. Additionally, we propose to train detectors in VIVID, and increase the accuracy with less real training photos. Our system can also connect to CarSim and simulate the response of real cars' suspension systems. (project home)
AnyCharge - An IoT-based Public Charging Service
AnyCharge is a public wireless charging service based on IoT technology. All the chargers are controllable by our cloud. We have deployed around 70 charging spots in Taiwan, Thailand, and Singapore (website, paper).
Mining Telecom Fraud using Social Network Analysis
Telecom fraud is one of the most prevalent crimes today and causes most property loss. To identify the roles of the fraudsters, we cooperate with Criminal Investigation Bureau of Taiwan and propose a Telecom Fraud Analysis Model (TFAM).
