Publications

A novel deep learning-based object detection framework, AirFormer, for Mars helicopter is proposed. AirFormer embeds a new feature-fusion attention module. In addition, we published a synthetic dataset from the viewpoint of the Mars helicopter ** SynMars-Air**, which refers to the data collected by the ZhuRong rover. The dataset is available at: https://github.com/CVIR-Lab/SynMars-Air.

Yifan Qi, Xueming Xiao, Meibao Yao, Yonggang Xiong, Lei Zhang, Hutao Cui

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 2024

a lightweight model with minimal computational complexity while maintaining high segmenting accuracy is proposed. We designed a lightweight squeeze window transformer module that focuses on window-scale feature extraction and is more effective in learning global and local contextual information

Yonggang Xiong, Xueming Xiao, Meibao Yao, Hutao Cui, Yuegang Fu

ISPRS Journal of Photogrammetry and Remote Sensing, 214,167-178,2024

A novel Transformer semantic segmentation framework for Martian rock images, consisting of an encoder-decoder structure connected through a Feature Enhancement Module (FEM) and a Window Transformer Block (WTB). We built a synthetic segmentatin dataset of Martian rocks, called SynMars.

Yonggang Xiongu, Xueming Xiao, Meibao Yao, Haiqaing Li, Hong Yang, Yuegang Fu

IEEE Transactions on Geoscience and Remote Sensing, 61,1-12,2023

The first U-shaped Transformer framework for Mars rock segmentation, consisting of a hierarchical encoder–decoder architecture with a feature refining module (FRM) connected between them.

Haiqaing Liu, Meibao Yao, Xueming Xiao, Yonggang Xiong

IEEE Transactions on Geoscience and Remote Sensing 61, 1-16, 2023

we propose a multi-object tracking framework for videos captured by UAVs, considering motion imperfection in the following two aspects. For motion blurring, this paper proposes a hybrid deblurring module that deals with the blurred frames while retaining the clear frames, trading off between video tracking performance and spatio-temporal consistency. For motion coupling, we proposed a motion compensation module to align adjacent frames by feature matching, and the corrected target position is obtained in the next frame to alleviate the interference of camera movement with tracking.

Song Cheng, Meibao Yao, Xueming Xiao

IEEE International Conference on Robotics and Automation (ICRA) (2023), ExCel London

A kernek-based rock detection framework is proposed for a planetary rover. Two algorithms are developed based on Kernel PCA and LRR. We also publised a small-scale Martian rock dataset called MarsData with labeled mask for rock segmentation.

Xiao, X., Yao, M., Liu, H., Wang, J., Zhang, L., & Fu, Y

IEEE Transactions on Neural Networks and Learning Systems (2021)

An innovative three dimensional hazard detection algorithm is proposed for a planetary safe landing mission. A novel robust scale estimator is proposed, which can achieve a breakdown point of more than 50 percentage.

Xueming Xiao, Hutao Cui, Yang Tian

IEEE Transactions on Aerospace and Electronic Systems 51, 4 (2015)

In this paper, we present a new autonomous rock detection approach through region contrast. New method deals with this issue in region level, which will significantly reduce the computational cost. Only low-level features are measured and a more reliable method is further proposed by introducing spatial factor and background similarity to the region contrast.

Xueming Xiao, Hutao Cui, Meibao Yao, Yang Tian

Advances in Space Research 60, 3 (2017)

This paper presents a high altitude hazard detection method for Mars landing. Towards safe landing requirements for future missions, hazard detection should be triggered as soon as possible. We therefore proposed a lidar based high altitude hazard detection approach, consisting of two main procedures, Terrain reconstruction and Hazard mapping. An interpolation based terrain reconstruction algorithm is proposed to model the datum surface of landing region.

Xueming Xiao, Meibao Yao, Hutao Cui, Yuegang Fu

Advances in Space Research 63, 8 (2019)

We propose a novel auto rock detection scheme via sparse based background modeling for planetary rover. Instead of pixel level image processing and pre training, we deal with the detection issue by region contrast enhancement. Our method consists of a rough detection algorithm to auto-label conservative background regions and a background modeling algorithm to precisely detect rocks. Background is reconstructed in feature space with the labeled regions as dictionary via sparse representation. Contrast map is enhanced by the error of original image and reconstructed background. Afterwards, rock is directly segmented by an auto threshold.

Xueming Xiao, Hutao Cui, Meibao Yao, Yuegang Fu, Wanqiang Qi

IEEE Congress on Evolutionary Computation (CEC). IEEE, 2018