Vignesh Ramakrishnan

• Customer Support, , Sample Application Development and Testing of CNN-Toolchain developed at Renesas to implement deep learning algorithms on embedded hardware • Support & Design of neural network architectures and sample applications on Deep Learning Frameworks to run efficiently on hardware • Managing and Guiding students • Continuous Integration using GitLab • Cross-compiling vision algorithms on Hardware • Integral role in specifying requirements for Renesas CNN-Toolchain

- 3D head pose estimation using PointClouds - 3D Animation of Driver Head - Practical experience in the form of test data collection for driver monitoring

- Motion Estimation was done in Inter Prediction mode using different methods like rate-distortion method and segment based method with Single Pixel Accuracy and Quarter Pixel Accuracy. - Peak Signal to Noise ratio and Sparsity was observed and compared. - Padding methods and Edge based methods were investigated to improve sparsity

Paper Published in SPIE Medical Imaging Conference: “Automated Detection of Schlemm's Canal in Spectral-Domain Optical Coherence Tomography” - Pre-processing: To remove noise from the given Optical Coherence Tomography (OCT) signal - Scale Invariant Feature Transform: Registration method to align the frames in the given video - 3D Segmentation: 3D Region Growing method implemented, Level Set method analysed - 3D Visualisation: Defect in OCT identified using Segmentation and visualised in 3D.

- To control and maintain electronic equipment in nuclear and thermal power plants. - To ensure safety of a thermal power plant.

- Different network models were implemented using a network simulation software called ns-3 based on C++. - Physical Layer Characteristics were altered and observed at the receiver.

Multimedia Communications Technology, Advanced Techniques in Signal Processing, Medical Image Processing, Acoustic Virtual Reality

Digital Signal Processing, Microprocessors, Antenna Engineering, Network protocols, Computer Architecture