Suren Sritharan

I am a Master's student at the Technical University of Munich, where I'm studying Computer Science with a focus on Machine Learning and Computer vision. I'm also working as a working student at AstraZeneca Computation Pathology on the image analysis of tumor cells, while completing my Thesis at the Computational Pathology Lab at TUM on Diffusion models for in-silico data generation. Prior to that, I was working as a working student at TUM, first in the Providentia++ project at the chair of Robotics, Artificial Intelligence and Embedded Systems, and then at the chair of Data Analytics and Machine Learning. I completed my Bachelor's degree at the University of Peradeniya with a B.Sc specializing in Computer Engineering.

Throughout my undergraduate studies, I was fortunate to be advised by many fantastic researchers. Soon after my bachelors, I had the opportunity to work with Prof. Janaka Ekanayake and Prof. Samath Dharmaratne on the potential of AI for COVID-19 related applications. Previously, for my final year research project, I worked with Prof. Roshan Ragel, Dr. Roshan Godaliyadda, Dr. Parakrama Ekanayake, and Dr. Vijitha Herath on image quality enhancement under low/high exposure conditions. During my internship at Nokia Bell Labs, I had the privilege of working with Prof. Haris Gačanin on the application of machine learning models for dynamic wireless networks.

Email  /  Github  /  Google Scholar  /  Linkedin  /  CV

• [Mar, 24] Our paper on Cooperative Perception got accepted for CVPR, 2024.
• [Apr, 23] I joined AstraZeneca Computational Pathology as a working student.
• [Oct, 22] I joined the DAML lab at TUM as a working student.
• [Dec, 21] I joined the Providentia++ as a working student
• [Oct, 21] I began my Master's in Informatik at the Technical University of Munich.

My primary research focuses on the intersection of machine learning and computer vision, particularly their applications in various domains. I am currently investigating the use of Diffusion models for synthetic data generation to improve downstream tasks such as cell segmentation. My research encompasses a broad range of topics within machine learning and machine perception, as demonstrated by the projects and publications listed below.

Proposed CMTCoop : A deep multi-model multi-view feature fusion model based on cross modal Transformers for 3D object detection. The model runs at near-real-time FPS and improves mAP over the current SoTA.

Semi-supervised / self-supervised learning methodology for 3D vehicular shape reconstruction based on BAAM, to pave the way for unsupervised techniques.

Dataset and model for cooperative perception in 3D vehicular object detection and tracking.

Asses the COVID-spreading risk levels from CCTV camera videos based on social distance measure, contact detection, mask identification, and temporal graph network to develop spread mitigation strategies for different environments.

A CNN model to solve the IID problem trained in an unsupervised manner using physics-based parameters derived from an image.

Review of the use of AI techniques for spatio-temporal modeling, forecasting, and impact modeling on diverse populations as it relates to COVID-19 and its potential for future applications.

Study of the practical limitations of learning models in a dynamic wireless environment.

Low light image enhancement utilizing paired and unpaired dataset using cyclic consistency.

Foreground estimation through background cancellation in dynamic background conditions based on classical signal processing techniques in spatio-temporal domain.

Sleep apnea identification in infants through a video feed.

Credits to Jon Barron.
Last updated Aug 2024.