Kamran Raza

Papers

Total Citations

H-Index

About

Kamran Raza is a computer vision researcher whose work focuses on advancing semantic segmentation techniques for challenging visual environments. His most cited paper, "Unified DeepLabV3+ for Semi-Dark Image Semantic Segmentation" (2022, 15 citations), tackles the critical problem of accurate visual perception under low-light conditions—a domain where traditional deep learning models often fail. By adapting the powerful DeepLabV3+ architecture for semi-dark imagery, Raza addresses unresolved challenges in automatic scene classification using predefined object classes, particularly when working with ResNet-based backbones. His contributions are significant for applications ranging from autonomous driving to surveillance systems that must operate reliably in non-ideal lighting. While his citation count reflects an emerging career, the specificity of his work on illumination-robust segmentation positions him at an important intersection of computer vision and practical deployment. Raza's research demonstrates how adapting established architectures for edge cases can push the boundaries of what deep convolutional neural networks can achieve in real-world visual perception tasks.