I'm a PhD student at Dartmouth working with Prof. Adithya Pediredla in the Rendering and Imaging Science Lab (RISc). My research focuses on building novel imaging systems by leveraging various sensor modalities (namely event cameras, RGBD sensors, etc.). My current research is mainly focused on Novel Acousto-Optic Imaging Systems for fast scanning and communication applications. I am particularly interested in exploring ways to harness ML or non-ML procedures for various real-world applications. Prior to my PhD, I completed my Master's working with Prof. PJ Narayanan at the intersection of 3D Vision and Real-Time Graphics.
Some things I'm interested in:
- Sensors (ToF, Motion Contrast, etc.)
- ML for Imaging
- Processing sensor data
- HPC and Automation for real-world applications
Some personal facts:
- I like capturing low-light images.
- Listening to music
- Driving (I enjoy driving long distances)
- I collect a lot of superhero merchandise and am fond of bubble tea.
Publications:
- Structured light with a million light planes per second - Dhawal Sirikonda, Praneeth Chakravarthula, Ioannis Gkioulekas, Adithya Pediredla. arXiv, 2025.
- GSN: Generalisable Segmentation in Neural Radiance Field - Vinayak Gupta, Rahul Goel, Dhawal Sirikonda, P. J. Narayanan. AAAI '24, 2024.
- FusedRF: Fusing Multiple Radiance Fields - Rahul Goel, Dhawal Sirikonda, Rajvi Shah, P. J. Narayanan. XRNeRF CVPR Workshop Paper, 2023.
- Interactive Segmentation of Radiance Fields - Rahul Goel*, Dhawal Sirikonda*, Saurabh Saini, P. J. Narayanan. CVPR '23, 2023.
- Real-Time Rendering of Arbitrary Surface Geometries using Learnt Transfer - Dhawal Sirikonda, Aakash KT, P. J. Narayanan. ICVGIP '22, 2022.
- Learnt Transfer for Surface Geometries - Dhawal Sirikonda, Aakash KT, P. J. Narayanan. HPG '22, 2022.
- StyleTRF: Stylizing Tensorial Radiance Fields - Rahul Goel*, Dhawal Sirikonda*, Saurabh Saini, P. J. Narayanan. ICVGIP '22, 2022.
- PRTT: Precomputed Radiance Transfer Textures - Dhawal Sirikonda, Aakash KT, P. J. Narayanan. arXiv, 2022.
- Transfer Textures for Fast Precomputed Radiance Transfer - Dhawal Sirikonda, Aakash KT, P. J. Narayanan. EuroGraphics '22, 2022.
- Neural View Synthesis with Appearance Editing from Unstructured Images - Pulkit Gera, Aakash KT, Dhawal Sirikonda, P. J. Narayanan. ICVGIP '21, 2021.
- Appearance Editing with Free-viewpoint Neural Rendering - Pulkit Gera, Aakash KT, Dhawal Sirikonda, Parikshit Sakurikar, P. J. Narayanan. arXiv, 2021.
(* = Equal contribution)
Projects:
- Using AO to Sculpt and Steer Light Planes at High Speeds for Scanning Applications: (Research Project) - This project focuses on leveraging Acousto-Optic (AO) techniques to dynamically sculpt and steer light planes at high speeds, enabling efficient and precise scanning applications.
- Underwater Communication: (Research Project) - This project explores the use of Acousto-Optic (AO) sculpting techniques to enhance underwater communication systems.
- Object retrieval from Radiance Fields: (Research Project in collaboration with Dr. Rajvi Shah) - Interactive object and sub-scene retrieval from scenes represented as Radiance Fields. The work involves growing high-confidence object content to encompass intricate details, aiming for accurate retrieval.
- Real-time rendering of Implicit Surfaces using Precomputed Radiance Transfer: (Thesis: CVIT, IIIT-H, 2022) - A simple yet fast approach to address the lack of storage schema in the functional representation of surfaces for the incorporation of Precomputed Radiance transfer (Spherical Harmonics) for both glossy and diffuse materials.
- Exploring storage schemas for Transfer Vector Storage: (Research Project, IIIT-H, 2022) - The project was based on the exploration of storage schemas (UV and Vertex attributes), to find optimal sampling and interpolation for artifact-free renders.
- Appearance Editing and Novel View Synthesis of captured data: (Research project: CVIT, IIIT-H, 2021) - The project extends Novel View synthesis pipelines to accommodate appearance edits. Preprocessing the data using Differentiable rendering for the separation appearance, followed by a disentangled rendering framework.