Dark Stereo Multi-Exposure Dataset (DSMD) for Extreme Low-Light Image Restoration and Depth Estimation
Depth estimation · Low-light imaging · Image restoration · Datasets · Stereo
We explore the emerging paradigm of computational imaging. Traditionally, imaging optics and image-based inference have been designed independently. By jointly optimizing optics and algorithms, we significantly improve imaging-system performance.
Latest research and publications from our lab.
↗NOVA-GS enables robust low-light 3D Gaussian Splatting through unified enhancement, self-supervised denoising, and noise-aware optimization, producing accurate geometry and consistent novel views.
DSMD is a public benchmark for extreme low-light stereo vision, providing multi-exposure stereo images with well-lit references. It supports joint evaluation of image enhancement and disparity estimation, advancing robust depth perception for autonomous driving, surveillance, AR/VR, and computational photography.









Stay informed about the latest news at IITM-CI-Lab
Depth estimation · Low-light imaging · Image restoration · Datasets · Stereo
Semantic segmentation · Thermal imaging · RGB-Thermal · Unsupervised
Single-photon imaging · 3D Reconstruction · Novel view synthesis · Computational imaging · Low-light imaging
Lensless imaging · Novel view synthesis · Computational imaging
Low-light imaging · Image restoration · Thermal imaging · RGB-Thermal
Generative models · Video editing · Faces
Event cameras · 3D Reconstruction · Novel view synthesis · Deblurring · Gaussian splatting
Underwater imaging · Image restoration · Novel view synthesis
Image restoration · Novel view synthesis · Computational imaging · 3D Reconstruction
HDR Imaging · 3D Reconstruction · Novel view synthesis · Gaussian splatting
Dual pixel sensor · Depth estimation · Computational imaging
Dual pixel sensor · Light-field imaging · Video · Knowledge distillation
Reflection removal · Semantic segmentation · Image restoration
Low-light imaging · Image restoration · Stereo
Inverse problems · Lensless imaging · Computational imaging
3D Reconstruction · Depth estimation · Lensless imaging · Computational imaging
Thermal imaging · Image restoration · Super-resolution
Light-field imaging · Video synthesis
Fourier ptychography · Computational imaging
Underwater imaging · Image restoration · Datasets
Compressive sensing · Video reconstruction · Coded exposure · Computational imaging
Low-light imaging · Image restoration · Light-field imaging
Low-light imaging · Image restoration · Real-time
Reflection removal · Image restoration · Real-time
Light-field imaging · Video · Self-supervised
Image restoration · Computational imaging · Low-light imaging
Thermal imaging · Super-resolution · Image restoration
Computational imaging · Computational photography · Privacy
Image forensics · Image compression · Machine learning
Inverse problems · Computational imaging · Compressive sensing · Generative models · Lensless imaging
Lensless imaging · Computational imaging
Depth estimation · Underwater imaging · Unsupervised
Event cameras · Machine learning · Action recognition
Explore our latest research talks, demonstrations, keynote presentations, and computational imaging breakthroughs.
The Computational Imaging Lab at IIT Madras builds the algorithms, optics, and learning systems behind next‑generation cameras ➙ spanning lensless and light‑field imaging, low‑light restoration, and neural rendering. We welcome postdoctoral researchers, graduate students, and undergraduates who want to push the boundaries of how machines see. Browse the opportunities below and reach out to Prof. Kaushik Mitra to apply.



For every track below, send your materials by email to Prof. Kaushik Mitra ➙ applications are reviewed on a rolling basis.
We are recruiting postdoctoral researchers and research scientists across several funded projects, looking for a PhD in computational imaging, computer vision, computational photography, optics, signal processing, machine learning, or a closely related area.
We welcome prospective PhD and MS by Research students interested in computational imaging, computer vision, computational photography, optics, signal processing, machine learning, or a closely related area. Admission and funding decisions for the IIT Madras EE graduate program are made centrally, so prospective students should first apply through the institute admissions office.
We're always glad to hear from IIT Madras undergraduate and dual‑degree students with strong math, signal processing, or coding skills.
Applications are reviewed on a rolling basis ➝ there is no fixed deadline, so reach out whenever you're ready.
The Computational Imaging Lab is located in the Electrical Sciences Block (ESB2), IIT Madras.
1st Floor, ESB2‑104, Electrical Sciences Block (ESB2), IIT Madras, Chennai, Tamil Nadu 600036
The Computational Imaging Lab welcomes applicants from all academic and personal backgrounds, and evaluates every candidate on the strength of their research interest and preparation.