At L2M Rail Private Limited, I develop and support the Machine Vision Inspection System for real-time railway safety inspection. I build and evaluate YOLO-based models and multi-stage YOLO and ResNet pipelines for detecting railway defects and classifying door and component conditions.
I also work with PaDiM anomaly detection and prepare railway image datasets from collection and annotation through augmentation, training, and evaluation. My work includes processing high-speed train images, GPU-based inference, model validation, and deployment preparation.
