Software Engineer AI at Nayan Technologies

NAYAN focuses on using cameras for Traffic Monitoring, Road-Safety as well as Monitoring Public Areas. In particular, the cameras are used for collecting visual data and providing video analytics and data with the use of people, computer vision, and artificial intelligence.

• Designed and developed re-training backend micro-services for C3 crowdsourced data annotation engine with automatic data train-validate-test and model retraining capability.

• Trained lane segmentation model used for Lane Occlusion classification with 91% accuracy validated on real feeds.

• Re-Trained Yolov4 model on a custom dataset for vehicle detection hence improving the model accuracy for our use case by almost 25%.

• Built and optimized License plate recognition models for both cloud and mobile phones.

• Develop 2D Keypoint localization using CNN for automatic pose estimation of cars.

• De-bugged and updated the logic for various traffic violation generation, which helped improve the generation of false violations by almost 40%.