Hi! I am Sri Siddarth Chakaravarthy (Pref. Sid / Sri Siddarth C), currently working as a Research Assistant in the Machine Learning and Vision Group at the Indian Institute of Technology Hyderabad (IITH), under the guidance of Dr. Vineeth N Balasubramanian.
My research revolves around Explainable AI, Continual Learning, and Multi-Modal Learning.
Previously, I collaborated with Dr. Suresh Sundaram at the Indian Institute of Science, extending our interactions from Nanyang Technological University, Singapore. I worked in the Artificial Intelligence and Robotics Laboratory (AIRL) on Motion Planning for ground robotics and 3D reconstruction. I also helped develop an AI-enabled autonomous drone for long range slant-angle detection using light-weight object detection models. (I continue to collaborate on projects remotely. Currently focused on 3D reconstruction)
I received my Bachelor's in Computer Science and Engineering from Vellore Institute of Technology (VIT)-Vellore, in 2022 (received the Research Excellence Award & Special Achiever’s Award). I conducted my undergraduate thesis under the guidance of Dr. Ashutosh Natraj and Dr. Vasantha W B.
At VIT, I've had the opportunity to intern at Hewlett Packard & Enterprises (with Manikanda Das R), Samsung R&D Institute (with Dr. Satya Kumar Vankayala), and Vidrona (with Dr.Ashutosh Natraj). Additionally, I conducted research at the VIT-Autonomous Research Center (ARC), contributing to lane detection models. In my final semester, I was honored to be a research exchange student (NTU-India Scholar) at Nanyang Technological University, Singapore, where I worked with Dr. Xie Ming on Restricted Coulomb Energy (RCE) Neural Networks for Semantic Segmentation in Autonomous Vehicles.
After graduation, I collaborated with OpenCV for Google Summer of Code as an open-source developer and contributed to the OpenCV model zoo.
My research interests lie in Continual Learning, ExplainableAI, and Multi-Modal Learning to improve generalization and overcome adversarial conditions in AI models. Recently, I have also developed an interest in computer graphics, and I am eager to expand my expertise into the realm of 3D computer vision.
During my free time I like to play the piano, play football, and paint pictures.
Publications
|
|
CLAM: Continual Learning with Multimodal ConceptsConference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2024 |
|
|
Safety and Reliability Integrated Physics-informed Neural Networks (PINN) for Obstacle AvoidanceIEEE International Conference on Electronics, Computing and Communication Technologies (CONECCT), 2024 |
|
|
RCE-Neural Network for Semantic Segmentation in Autonomous VehiclesArtificial Intelligence for Autonomous Robots (MDPI) *Abstract Accepted* |
Presentations & Symposiums
|
Design of Farmer friendly interface using kiosk8th International Conference on Research into Design (ICORD), Jan 2021 |
|
Optical Character Recognition using CNN with Air-writing for Indian LanguageComputational Intelligence Issues in Blockchain, AI, and ML, May 2021 |
News
Select Projects
|
|
Predictive and Prescriptive Analysis for Power Transmission Components.Sri Siddarth Chakaravarthy* Bachelor's Thesis Developed models for identifying the health of power transmission components in macro and microgrid systems. Monitored rust-levels in components, identified sag in jumper cable lines, and fault diagnosis in components. Incorporated Augmented Reality for viewing insulator components as 3D objects. |
|
|
Microservice Orchestration using Netflix Conductor engine for E-Commerce application.Work @ Hewlett-Packard Enterprises Developed microservices for an e-commerce application by containerization of services on docker and modelled the Netflix conductor engine to enable microservice orchestration. Slide |
|
|
Optimal BS Transmit Power for 5G and 6G Systems using MLWork @ Samsung Research & Development Institute Leveraged the use of Deep Learning to realize a mapping between User Equipment(s) configurations and the power allocation policies within 5G Base Stations. Simulated the environment using Matlab 5G toolkit. |
|
|
Real-time Tamil Character Recognition.Sri Siddarth Chakaravarthy* Worked on evaluating the performance of ConvLSTM for estimating Optical flow in autonomous vehicles using the Cityscapes dataset. This was done for preparing the input for the problem of driver maneuver prediction. Code Slide |
|
|
3D Rendering using OpenGL for Lighting Manipulation and Path TracingSri Siddarth Chakaravarthy* Developed a simple OpenGL application that incorporated basic idea of path tracing on sphere objects with a single light source. Included variations in Phong lighting model to show variations in Diffue and Spectral properties. Code Slide |
|
|
Optical Flow Estimation in autonomous Vehicles using ConvLSTMs.Sri Siddarth Chakaravarthy* Worked on evaluating the performance of ConvLSTM for estimating Optical flow in autonomous vehicles using the Cityscapes dataset. This was done for preparing the input for the problem of driver maneuver prediction. |
|
Model Quantization for Light-weight Object Detection on the Edge.Work @ OpenCV (Google Summer of Code'22) Worked on model quantization for NanoDet and YoloX with optimized performance on resource-restricted devices. |
|
Mesh Extraction from InfoNerf.Sri Siddarth Chakaravarthy* Working on a mesh extraction pipeline for 3D reconstruction models. |
Honors 🏆
- Awarded with Excellence Award for outstanding work project development & presentation at VIT-Open house. ⭐
- Received Special Achiever’s Award for outstanding research work during Bachelor's. ⭐
- Selected for the HPE-CTY to conduct research with Hewlett Packard Enterprises’s research & development team. ⭐
- Selected as a Samsung-PRISM Research collaborator. ⭐
- Recipient of NTU-India Connect Exchange Scholar, selected as one of the research exchange students from a competitive pool of candidates. ⭐
- Top 1% in branch in the passing academic year. ⭐