Xinhao Xiang

Xinhao Xiang

Ph.D. student in Computer Science

UC Davis

About Me

I am currently a student pursing Doctorate of Philosophy in Computer Science degree in University of California, Davis advised by Dr. Jiawei Zhang.

My research involve in Computer Vision Perception (3D Object Detection, Point Cloud Perception, Segmentation, Facial Landmark Detection, Multi-model Fusion, Open-vocabulary Detection, Efficient Vision Models, Video Understanding), Generation & Reconstruction Understanding (Generated Video Quality Assessment, Multi-modal Foundation Models, 3D Reconstruction Acceleration), and AI Robotics (Swarm Intelligence, Path Planning, Multi-target Optimization).

Interests
  • Computer Vision Perception
  • Generation & Reconstruction Understanding
  • AI Robotics
Education

  • Ph.D. in Computer Science, 2022-present

    UC Davis

  • M.S. in Computer Science, 2021-2022

    UC Davis

  • B.E. in Computer Science and Technology, 2017-2021

    Southern University of Science and Technology (SUSTech)

Publications

Are AI-Generated Driving Videos Ready for Autonomous Driving? A Diagnostic Evaluation Framework
Created a diagnostic evaluation framework for AI-generated driving videos: a failure modes taxonomy, a AI driving videos benchmark, and a driving-aware evaluator combining static/temporal cues.
Xinhao Xiang, Abhijeet Rastogi, Jiawei Zhang
Preprint
PDF Cite
Are AI-Generated Driving Videos Ready for Autonomous Driving? A Diagnostic Evaluation Framework
AIGVE-Tool: AI-Generated Video Evaluation Toolkit with Multifaceted Benchmark
AIGVE-Tool introduces a unified, modular framework, a benchmark (AIGVE-Bench) for comprehensive, standardized evaluation of AI-generated videos across multiple quality dimensions, and a structured five-category taxonomy of evaluation metrics.
Xinhao Xiang, Xiao Liu, Zizhong Li, Zhuosheng Liu, Jiawei Zhang
The 21th International Conference on Computer Vision Theory and Applications (VISAPP), 2026
PDF Cite Code Dataset Project Source Document
AIGVE-Tool: AI-Generated Video Evaluation Toolkit with Multifaceted Benchmark
Open-Vocabulary Multimodal 3D Object Detection with Attributes
A novel open-vocabulary multi-modal 3D object detection method which can incorporate multi-view input and detect complex events (including attributes) without needing novel class anchor size; proposed a dataset for open-vocabulary attribute detection
Xinhao Xiang, Kuan-Chuan Peng, Suhas Lohit, Michael Jeffrey Jones, Jiawei Zhang
The British Machine Vision Conference (BMVC), 2025 Oral
PDF Cite Dataset Poster Slides Video
Open-Vocabulary Multimodal 3D Object Detection with Attributes
A Survey of AI-Generated Video Evaluation
A comprehensive review of current methodologies of AI-Generated Video Evaluation (AIGVE), highlighting emerging field, and guiding for future research directions.
Xiao Liu, Xinhao Xiang, Zizhong Li, Yongheng Wang, Zhuoheng Li, Zhuosheng Liu, Weidi Zhang, Weiqi Ye, Jiawei Zhang
Preprint
PDF Cite
A Survey of AI-Generated Video Evaluation
EffiPerception: an Efficient Framework for Various Perception Tasks
A general framework with efficient feature extractions and layers (Sparse Down-Sampling, Global Spatial Aggregation). It achieves great accuracy-speed performance with relatively low memory cost under several perception tasks (including 3D/2D detection/segmentation).
Xinhao Xiang, Simon Dräger, Jiawei Zhang
The Pacific Rim International Conference Series on Artificial Intelligence (PRICAI), 2025
PDF Cite Poster
EffiPerception: an Efficient Framework for Various Perception Tasks
3DifFusionDet: Diffusion Model for 3D Object Detection with Robust LiDAR-Camera Fusion
A novel 3D Object Detection method that structure the task as a denoising diffusion process from noisy 3D boxes to target boxes.
Xinhao Xiang, Simon Dräger, Jiawei Zhang
IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2026
PDF Cite
3DifFusionDet: Diffusion Model for 3D Object Detection with Robust LiDAR-Camera Fusion
FusionViT: Hierarchical 3D Object Detection via Lidar and Camera Vision Transformer Fusion
A ViT-pured hierarchical architecture that embed both Images and Point Clouds, fused via a mixViT for efficient 3D Object Detection.
Xinhao Xiang, Jiawei Zhang
The 18th International Conference on Machine Vision (ICMV), 2025 Oral
PDF Cite Slides
FusionViT: Hierarchical 3D Object Detection via Lidar and Camera Vision Transformer Fusion
Robotic Brain Storm Optimization: A Multi-target Collaborative Searching Paradigm for Swarm Robotics
A BSO-based collaborative searching framework for swarm robotics.
Jian Yang, Donghui Zhao, Xinhao Xiang, and Yuhui Shi
The Twelfth International Conference on Swarm Intelligence (ICSI), 2021 Oral
PDF Cite DOI
Robotic Brain Storm Optimization: A Multi-target Collaborative Searching Paradigm for Swarm Robotics
Exploration Enhanced RPSO for Collaborative Multitarget Searching of Robotic Swarms
An multitarget searching method for robotic swarms.
Jian Yang, Ruilin Xiong, Xinhao Xiang, and Yuhui Shi
Complexity, Hindawi, 2020
PDF Cite Code Video DOI
Exploration Enhanced RPSO for Collaborative Multitarget Searching of Robotic Swarms

Archives

Collaborative Multitarget Searching of Robotic Swarms Based on PF and BSO Algorithm
[Undergraduate thesis] Proposed a novel strategy of collaborative motion and multi-target searching for swarm intelligent robots under decentralized topology architecture with limited sense of perception and infocommunication ability;
Xinhao Xiang
Jul 1, 2021
PDF Cite
Collaborative Multitarget Searching of Robotic Swarms Based on PF and BSO Algorithm
Application Level Anomaly Detection in Autonomous Vehicles
[UCInspire research intern technical report] Employed the Carla simulator to perform real-time vehicle driving simulation. Collected and extracted various sensors’ data. Proposed a new strategy to detect outlier situations for this settings.
Xinhao Xiang
Nov 29, 2020
PDF Cite
Application Level Anomaly Detection in Autonomous Vehicles

Contact