At 深圳市未来眼科技有限公司, I optimize stereo-camera calibration workflows and software, including dynamic adjustment of camera parameters during data collection. I also work on data processing and performance optimization for two-hand keypoint detection in first-person Ego scenes.
At 广州斯睿特智能科技有限公司, I worked on AI inspection projects for automotive body defects, textile flaws, and assembly behavior. For a vehicle-body inspection project, I trained and deployed vision models and iterated on production-line data, reducing the reported false-positive rate from about 10% to 0.1% and reaching zero missed detections.
At 广东省农业科学院作物所, I built a cotton disease-and-pest RAG question-answering system combining LangChain, a local Ollama model, and YOLOv8 disease detection. I’ve also worked on cotton-growth prediction and field segmentation, and developed an Android-based lychee leaf disease detection system with offline inference.
