R&D Machine Learning/ Computer Vision Engineer
Current- Specialized in real-time computer vision for hotspot detection on solar panels using thermal camera technology and a first principles approach. Developed efficient detection algorithms on Raspberry Pi 4 using C.
- Designed and implemented an obstacle avoidance algorithm to enable autonomous robots to detect and navigate around objects in real time, improving safety and operational efficiency.
- … Show more
- Spearheaded the development of image annotation processes, converting annotations to JSON format for training object detection models with TensorFlow and PyTorch, enhancing model training efficiency and accuracy.
- Developed a comprehensive analysis toolkit for evaluating object detection accuracy, which includes generating confusion matrices and monitoring CPU and memory usage to optimize system performance.
- Created and deployed a semi-supervised learning system for real-time model training; the system annotates detected objects, generates JSON files automatically, and refines learning processes continuously.