Ghost-Valley AI Technology (Shenzhen) Co., Ltd (Nov 2023 – Jan 2024 & Sep 2024 – Present)

• Contributed to the development of a subway tunnel obstacle detection and ranging system, using binocular vision and advanced AI models.
• Annotated 850 images of tunnel obstacles, creating a comprehensive dataset for training machine learning models.
• Implemented and trained a YOLOv8 segmentation model for effective obstacle detection.
• Developed a C++ solution integrating the ELAS (Efficient Large-Scale Stereo Matching) algorithm with a ZED 2 stereo camera, generating real-time disparity maps at 12 frames per second on high-resolution (1344x376p) images.
• Encapsulated the ELAS algorithm into shared libraries (.dll/.so) for cross-platform deployment and integration.

Disparity Map Generated with ELAS and ZED 2 Camera.

Shenzhen Academy of Disaster Prevention and Reduction (Jul 2024 – Aug 2024)

• Assisted in experiments of imaging technology based on distributed acoustic sensing (DAS) fiber and background noise.
• Set up over 4,000 meters of DAS fiber, assisted in fiber splicing, and contributed to background noise data collection.
• Participated in the compilation of the detection report for the seismic safety detection and alarm system of the China Spallation Neutron Source, focusing on vibration data visualization and anomaly analysis using MATLAB.