Research

My research areas include

• Computer vision (e.g., object tracking, human pose estimation, human behavior prediction)
• Deep learning (e.g., multi-modal learning, self-supervised learning, graph-based learning)
• Image fusion (visible-infrared, multi-focus, multi-exposure)
• Multimodal applications (e.g., RGB-T tracking, RGB-T segmentation, RGB-T crowd counting)
• Robotics and AI (Multimodal robot perception)
• Ethical AI (pedestrian privacy protection)

Some of my projects are as follows:

1. Deep Learning-based image fusion: methods, benchmarks, and multi-modal applications

Related publications:

1. X. Zhang, Y. Demiris. Visible and Infrared Image Fusion using Deep Learning, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 8, pp. 10535-10554, 2023. (ESI Highly Cited Paper)
2. X. Zhang. Deep Learning-based Multi-focus Image Fusion: A Survey and A Comparative Study, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, No. 9, pp. 4819 – 4838, 2022. [Link] (ESI Highly Cited Paper)
3. X. Zhang. Benchmarking and Comparing Multi-exposure Image Fusion Algorithms. Information Fusion, vol. 74, pp. 111-131, 2021. (The first multi-exposure image fusion benchmark) [Benchmark link]
4. X. Zhang, P. Ye, G. Xiao. VIFB: A Visible and Infrared Image Fusion Benchmark, In the Proceedings of IEEE/CVF Conference on Computer Vision Workshops, 2020. (The first image fusion benchmark, which has been utilized by researchers from more than 10 countries.) [Benchmark link]
5. X. Zhang, P. Ye, S. Peng, J. Liu, G. Xiao. DSiamMFT: An RGB-T fusion tracking method via dynamic Siamese networks using multi-layer feature fusion. Signal Processing: Image Communication, vol. 84, 2020.
6. X. Zhang, P. Ye, D. Qiao, J. Zhao, S. Peng, G. Xiao. Object Fusion Tracking Based on Visible and Infrared Images Using Fully Convolutional Siamese Networks. In Proceedings of the 22nd International Conference on Information Fusion, 2019.

2. Pedestrian-centric visual computing

(1) Pedestrian trajectory prediction based on graph neural network

(2) Pedestrian crossing intention prediction based on graph neural network

(3) Pedestrian tracking

Related publications:

1. X. Zhang*, P. Angeloudis, Y. Demiris. Dual-branch Spatio-Temporal Graph Neural Networks for Pedestrian Trajectory Prediction, Pattern Recognition, vol. 142, 2023.
2. X. Zhang*, P. Angeloudis, Y. Demiris. ST CrossingPose: A Spatial-Temporal Graph Convolutional Network for Skeleton-based Pedestrian Crossing Intention Prediction, IEEE Transactions on Intelligent Transportations, vol. 23, no. 11, pp. 20773-20782, 2022.

3. Responsible and ethical AI: Pedestrian privacy protection

Many videos are captured to train AI models. We aim to protect pedestrian privacy in videos captured by cameras mounted on robots and vehicles while maintaining the utility of the anonymized videos.

Related publications:

1. Z. Zhao, X. Zhang*, Y. Demiris. 3PFS: Protecting pedestrian privacy through face swapping, IEEE Transactions on Intelligent Transportations, 2024.

4. RGB-based Computer vision

1. J. Liu, P. Ye. X. Zhang*, G. Xiao. Real-time long-term tracking with reliability assessment and object recovery. IET Image Processing, vol. 15, no. 4, pp. 918-935, 2021.
2. J. Liu, G. Xiao, X. Zhang*, P. Ye, X. Xiong, S. Peng. Anti-occlusion object tracking based on correlation filter. Signal, Image and Video Processing, vol. 14, no. 4, pp. 753-761, 2020.
3. J. Zhao, G. Xiao, X. Zhang*, D. P. Bavirisetti. An improved long-term correlation tracking method with occlusion handling. Chinese Optics Letters, vol. 17, no. 3, pp. 031001-1: 031001-6, 2019.

5. Robotics and AI, especially Multimodal robot perception

I am collaborating with the Personal Robotics Lab at Imperial College London. We are aiming at improving the perception capability of mobile robots.

6. AI for Science

An ongoing collaborative project.