Ebin Zacharias

Topic: Deep Learning based Scene Text Recognition in Mobile Outdoor Application Grade: 1.0 - Proposed a scene text recognition system following a two-stage pipeline. (System Accuracy = 98.2%) - Text detection was achieved using a Connectionist Text Proposal Network-based approach. (Accuracy = 100%) - Analysed attention mechanism and text recognition using LSTM network was trained. (Accuracy = 98.92%) - A synthetic data generator is proposed using Python and OpenCV.

Developed a scene-text detection system using image processing techniques.

- Developed a pipeline for swap-body detection and recognition by extracting the ILU code. Gained experience with data acquisition and image annotation.

- Related Coursework: 2D Image Processing, Foundations of Software Engineering, Software Development for Commercial Vehicles - Seminar on "Fuel Consumption Control Strategy for A Parallel Mild Hybrid Electric Vehicle" - Seminar on "Positive Temperature Coefficient Heaters (PTC)" - Scientific Writing - "Electrification of Trucks: Current Limitations and the Impact of Electric Drive"