Enhanced License Plate Recognition Using GANs & Deep Learning

License-plate recognition becomes significantly harder when plates are small, blurred, compressed, or captured under difficult imaging conditions.

Developed an end-to-end deep-learning framework for license-plate detection and recognition and investigated multiple detection architectures including CNN-based methods, SSD, Faster R-CNN, and YOLO-family approaches.

Integrated adversarial super-resolution techniques based on SRGAN to enhance plate-image quality before downstream recognition.

Experimented with SRGAN architecture/loss modifications and image-processing pipelines as part of the broader license-plate research work.