Ms. Varsha Singh | Deep Learning for Computer Vision | Best Researcher Award
Research Scholar (Ph.D.) | National Institute of Technology | India
Ms. Varsha Singh is a dedicated researcher at the National Institute of Technology, Tiruchirappalli, specializing in deep learning, computer vision, and efficient image super-resolution architectures. Her research is centered on developing lightweight yet high-performing neural models that enhance perceptual image quality through advanced multi-scale feature extraction, attention mechanisms, and dense connectivity designs.Her notable contribution, Optimized and Deep Cross Dense Skip Connected Network for Single Image Super-Resolution (DCDSCN) published in SN Computer Science introduced a cross-dense skip-connected framework that effectively balances computational efficiency and reconstruction accuracy. The proposed Cross Dense-in-Dense Convolution Block (CDDCB) leverages multi-branch feature fusion and short-path gradient propagation, achieving superior PSNR and SSIM performance across benchmark datasets such as Set5, Set14, BSD100, and Urban100. Building on this foundation, her subsequent work Multi-Scale Attention Residual Convolution Neural Network for Single Image Super-Resolution (MSARCNN) published in Digital Signal Processing Elsevier advances the field through the integration of Squeeze-and-Excitation and Pixel Attention modules within a multi-scale residual framework, enabling fine-grained texture recovery while maintaining low model complexity.With two international journal publications, Ms. Singh’s work demonstrates a strong emphasis on hierarchical feature fusion, adaptive attention modeling, and efficient neural design for real-time visual intelligence. She actively contributes to the scholarly community as a reviewer for the International Research Journal of Multidisciplinary Technovation (Scopus Indexed), where she has evaluated research papers in deep learning and image processing.Ms. Singh’s contributions bridge theoretical innovation and practical deployment, particularly in resource-constrained imaging and enhancement systems, fostering advancements in next-generation super-resolution and perceptual image restoration. Her research continues to strengthen the global discourse on AI-driven visual computing, supporting the development of intelligent and sustainable imaging solutions for diverse real-world applications.
Featured Publications
1.Singh, V., Vedhamuru, N., Malmathanraj, R., & Palanisamy, P. (2025). Multi-scale attention residual convolution neural network for single image super-resolution (MSARCNN). Digital Signal Processing, 146, 105614.
2.Singh, V., Vedhamuru, N., Malmathanraj, R., & Palanisamy, P. (2025). Optimized and deep cross dense skip connected network for single image super-resolution (DCDSCN). SN Computer Science, 6(5), 495.
Ms. Varsha Singh’s research advances efficient deep learning and image super-resolution, enabling high-quality visual reconstruction with minimal computational cost. Her innovations contribute to scientific progress in AI-driven imaging, with potential applications in medical diagnostics, remote sensing, and real-time visual enhancement, driving global innovation in sustainable and intelligent vision technologies.