Prashant Laddha
Papers
2
Total Citations
11
H-Index
1
About
Prashant Laddha is a researcher advancing the frontier of 3D scene understanding, with a focus on autonomous driving, robotics, and augmented/virtual reality. His work tackles the core challenges of processing sparse, unstructured point cloud data through innovative deep learning architectures. Laddha’s key contribution, "Segment-Fusion: Hierarchical Context Fusion for Robust 3D Semantic Segmentation" (2022, 10 citations), directly addresses the persistent "part-misclassification" problem—where models incorrectly label segments of the same object—by introducing a hierarchical context fusion mechanism that significantly improves segmentation accuracy. This work is a building block for reliable perception in real-world systems. More recently, in "Ace-of-Spades: Accelerating Spatially Sparse Convolution for 3D Scene Understanding" (2025, 1 citation), Laddha targets the computational bottleneck of 3D CNNs on point clouds, proposing a method to dramatically speed up spatially sparse convolutions. This efficiency-focused research is critical for deploying high-performance 3D models on resource-constrained platforms like autonomous vehicles. Laddha’s work sits at the intersection of algorithmic robustness and practical acceleration, making him a notable voice in the push toward real-time, reliable 3D perception.
Research Focus
Key Achievements
Top Papers
- 1
- 2