Justin Liang

Papers

Total Citations

H-Index

About

Justin Liang is a computer vision researcher whose work focuses on the fundamental challenge of reconstructing high-quality 3D objects from sparse, partial observations—a critical capability for applications in robotics, graphics, and autonomous driving. His research addresses a key gap in neural implicit modeling: while these methods excel on synthetic or dense datasets, they consistently fail when applied to real-world, single-view data. Liang’s major contribution lies in diagnosing and repairing these failures. In his most cited work, “Mending Neural Implicit Modeling for 3D Vehicle Reconstruction in the Wild” (2022, 24 citations), he developed techniques to recover high-fidelity 3D vehicle shapes from the sparse, partial views typical of street-level imagery. His earlier paper, “Secrets of 3D Implicit Object Shape Reconstruction in the Wild” (2021, 8 citations), laid the groundwork by systematically analyzing why neural implicit models struggle in uncontrolled environments. Liang’s work is notable for bridging the gap between academic benchmarks and practical deployment, making 3D reconstruction robust enough for real-world perception systems. His research is essential reading for anyone working on 3D vision for autonomous vehicles or robotics in unstructured environments.