Swapnil Sahu

I've built computer vision systems at Logic Fruit Technologies for aerial object classification, safety compliance detection, and space object recognition. My drone-versus-bird classification model reached 95% accuracy using optimized five-second trajectory intervals.

I developed a YOLO-based PPE detection system for Agilex 7, including an unsupervised and semi-supervised annotation pipeline that achieved approximately 80% accuracy. I also generated synthetic ISAR images from 3D mesh models to improve recognition of satellites and debris where real radar data was limited.

Beyond internship work, I've built image super-resolution and human action recognition systems using EDSR, ESRGAN, CNN-LSTM architectures, PyTorch, TensorRT, ONNX, TensorFlow Lite, OpenCV, and Flask.

I'm pursuing a B.Tech in Computer Science Engineering with Artificial Intelligence and Machine Learning at Graphic Era University, and hold certifications in Azure AI Fundamentals, AWS Certified AI Practitioner, and Google Cloud Fundamentals.