Education:
- Master’s in Geomatics Engineering
- Tongji University
- Specialization: Computer Vision and Remote Sensing
- Research focus: Object Detection, Semantic Segmentation
- Bachelor’s in Geomatics Engineering
- Tongji University
- Relevant courses: Photogrammetry, Digital Image Processing, Communication Principles, GIS Development, Numerical Computing, Error Propagation and Adjustment, Data Structures and Databases, C++ Programming
Project Experience:
Semantic Segmentation of Underground Parking Slot using TDA4 (Team Lead)
- Led a project involving open-source dataset collection, data preprocessing, and training semantic segmentation models (UNet, Swin-Transformer, DeeplabV3)
- Conducted channel-level compression on the UNet network and tested models of various sizes.
- Converted the model format from PyTorch to ONNX to TIDL.
- Deployed the model on the TDA4 platform, performed INT 8 quantization compression, and conducted performance testing with models of different sizes.
- Future work involves collecting a semantic segmentation dataset using a UGV platform for underground parking HD mapping (Multi-sensor fusion).
Pedestrian Tracking, Counting, and Alert System using Deep Learning (Team Lead)
- Implemented pedestrian detection, tracking, line-crossing detection, and counting using YOLOv5 and Deepsort.
- Utilized serial communication to control alert devices when pedestrian flow exceeded a threshold.
- Deployed the system on the Jetson Nano platform and used TenserRT for inference acceleration（3x speedup）.
Intelligent Analysis of Tower Crane Top-view Monitoring Videos (Team Lead)
- Detected and Classified tower crane loads and ground materials.
- Fused data from tower crane motion sensors and gravity sensors to record load segments, generate daily load reports, stitching panoramic map of the construction site and reconstruct a 3D view of the construction site.
- Explored traditional feature matching frame difference methods and implemented rotation box object detection using YOLOv5.
- Using ONNXRUNTIME for model deployment(C++)
Bare Soil Recognition using a Boosted Cascade of CNN and Manual Features(Team Lead)
- Used semantic segmentation model(ResNet50+UPerNet) and manual features(Contifidence, RGB, Hue) to identify the proportion of bare soil in construction site images.
- Implemented the program using Python and C++ (OpenCV).
Unmanned Aerial Vehicle Pose Measurement Using High-Speed Camera(Undergraduate Thesis)
- Conducted camera calibration for internal orientation elements.
- Used a total station to measure control points for external orientation elements.
- Analyzed aircraft position and pose information using Photogrammetry and visualize data using Matlab.
Internship Experience:
- ShiZhuang Information Technology Co., Ltd.
- Image Algorithm Intern
- Optimized object detection models for the Image Search project.
- Assisted in data cleaning, labeling, and evaluating model accuracy.
- AIGC diffusion model research and image2image model generating test.
- China Aerospace Science & Industry Corporation（CASIC）
- Software Testing Intern
- Debugged navigation positioning software code (C++).
- Wrote and verified data formatting programs.
Personal Strengths:
Engineering Skills:
- Proficient in C++ and Python; experience with VB, C#, Lisp, Matlab, and WeChat Mini Program development.
- Completed end-to-end deep learning tasks, from data annotation to model deployment.
- Familiar with different operating systems (Windows, Linux, macOS) and programming tools (Git, Vim, Docker, ssh, CMake).
- Understanding of computer principles and basic knowledge of network communication.
Language Proficiency:
- IELTS score of 7.0 in English, N2 level in Japanese.
- Basic proficiency in German and Cantonese.
