PhD candidate · James Cook University
I build deep-learning perception for autonomous driving, generative models for inverse problems, and the systems that put them to work in the field — from radar signal chains and data pipelines to a crocodile-detection system for the Queensland Government.
Four sides of my work: perception for autonomous driving, generative models, applied AI systems, and the signals and hardware underneath.
Experience & skills
Deep learning for 3D detection, radar–LiDAR fusion and tracking, from research models to production pipelines.
Diffusion priors and self-diffusion for recovering clean images and point clouds from degraded, incomplete data.
Taking models into the field and building the systems around them — data loops, deployment, and full products.
Radar signal processing, multi-sensor synchronisation and embedded hardware — the layer the models stand on.
Y. Yang, G. Luo
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · pp. 26711–26720
Zero-shot, self-supervised blind imaging: an iterative reverse self-diffusion process that starts from pure noise and jointly refines the sharp image and the unknown blur kernel — no pre-training.
PaperarXivProjectY. Yang, J. Liu, G. Luo, H. Li, E. Ahn, M. Rahimi Azghadi, T. Huang
arXiv:2505.09887
Enhances sparse, noisy radar point clouds toward LiDAR-like density with a diffusion prior learned from arbitrary, unpaired LiDAR data.
arXivCodeProjectY. Yang*, J. Liu*, T. Huang, Q.-L. Han, G. Ma, B. Zhu
IEEE Transactions on Circuits and Systems for Video Technology · 35(5):4130–4143 · *equal contribution
Fuses radar range–azimuth heatmaps with LiDAR in bird's-eye view for all-weather detection: +13.1% (clear) and +19.0% (fog) at IoU 0.8.
IEEEarXivCodeProjectRecent
I work across the whole stack of AI for perception. At Vanjee Technology in Beijing I spent four years as an algorithm engineer on autonomous and vehicle–road cooperative driving: 3D object detection and radar–LiDAR fusion, the MIMO radar signal chain underneath them, and a data closed-loop system I led to keep models improving after deployment — with nine granted patents along the way.
A public Chinese-language data site covering ~490 occupations, rebuilt automatically every month.
Computer vision that tells crocodiles apart from other floating objects, for real-time public-safety monitoring.
The full processing chain of an in-house millimetre-wave radar toolkit, from raw array signals to detections.