Abdelrahman Eldesokey

Papers

Total Citations

296

H-Index

About

Abdelrahman Eldesokey is a leading researcher in computer vision, specializing in deep learning for sparse and irregularly spaced data—a critical challenge for autonomous driving, robotics, and surveillance. His major contribution is pioneering the concept of confidence propagation through convolutional neural networks (CNNs), enabling robust regression on sparse depth maps. His seminal 2019 paper, "Confidence Propagation through CNNs for Guided Sparse Depth Regression," has garnered 221 citations, establishing a foundational method for handling incomplete sensor data. This work builds on his earlier 2018 paper (75 citations), which first introduced the idea of propagating confidences through CNNs for sparse data regression. Eldesokey’s innovations allow CNNs to effectively process inputs from LiDAR and other sparse sensors, directly impacting real-world systems that require accurate depth perception under challenging conditions. His research bridges the gap between dense image processing and sparse sensor data, offering practical solutions for scene understanding. With a growing citation record and clear applications in safety-critical domains, Eldesokey is recognized for transforming how neural networks handle irregular data, making him a key figure in advancing robust perception for autonomous systems.