Zihao Wan

Papers

Total Citations

H-Index

About

Zihao Wan is a rising researcher in 3D computer vision, with a focused expertise in instance segmentation for cluttered and complex environments. His work addresses a critical challenge: how to accurately identify and separate individual objects in 3D point clouds when objects are densely packed or overlapping. Wan’s major contribution is the development of novel neural network architectures that go beyond traditional geometric proximity. In his paper "IAN: Instance-Augmented Net for 3D Instance Segmentation" (2023, 3 citations), he introduced a method that intelligently identifies whether neighboring points belong to the same object, filtering out noisy features that degrade performance. Building on this, his "IGN: Instance-Guided Net for 3-D Instance Segmentation in Cluttered Scenes via Monocular Depth Sensor" (2024, 3 citations) specifically targets the under-explored problem of segmentation in highly cluttered scenes—a scenario common in real-world robotics but underrepresented in standard datasets. Though early in his career, Wan’s work is directly applicable to robot manipulation and sensor data processing, offering practical solutions for machines to perceive and interact with messy, real-world environments. His research bridges a critical gap between theoretical segmentation models and the demands of autonomous systems operating in unstructured spaces.