Computer vision algorithm for transformation of Night time Images to corresponding Day time
Project Summary
“Computer vision algorithm for transformation of Night time Images to corresponding Day time.”
RESEARCH GRANTRs. 9,94,9432020 – 2021
RESEARCH OVERVIEW & DETAILS
Project Scope & Details
Computer vision algorithm for transformation of Night time Images to corresponding Day time.
ARCHITECTURAL & SCIENTIFIC HIGHLIGHTS
01.
Retinex-Grounded CycleGAN Architecture
Physically decouples scene illumination from reflectance to restore natural daytime colors from low-light photography.
02.
Perceptual Contrast Preservation Loss
Prevents overexposure blooming around localized light sources while brightening deep underexposed shadows.
03.
Structural Similarity & Edge Retention
Enforces edge-aware gradient constraints, preserving fine road surface and architectural textures.
04.
Real-Time Embedded Edge Inference
Quantized network architectures optimized for deployment on in-vehicle intelligent transportation systems.
ACADEMIC PUBLICATIONS [1]
2022Proceedings of the Satellite Workshops of ICVGIP 2021
Semantic Map Injected GAN Training for Image-to-Image Translation
Authors: Balaram Singh Kshatriya, Shiv Ram Dubey, Himangshu Sarma, Kunal Chaudhary, Meva Ram Gurjar, Rahul Rai, Sunny Manchanda
RESEARCH TEAM
- Dr. Himangshu Sarma
EXTERNAL COLLABORATORS
- ▸DRDO Young Scientist Laboratory – Artificial Intelligence (DYSL-AI)
FUNDING & GRANTS
SANCTIONED BUDGETRs. 9,94,943
SPONSORING AGENCYDRDO Young Scientist Laboratory – Artificial Intelligence (DYSL-AI), DRDO, Govt of India
PROJECT TIMELINE2020-05-01 – 2021-05-01
TECHNOLOGIES
Computer vision