Computer Vision Publications

In the AccessibleMaps project, we investigate the detection of accessibility features in buildings using a camera or other sensors by developing new methods based on artificial intelligence. Many of these methods can be used both indoors and outdoors.

Various mobile prototypes are being developed, tested and evaluated. Since people with blindness work in the project, the use of these prototypes is also considered for this group.

The following aspects will be investigated in the work:

1. Development of robust methods to interpret a scene, i.e., correctly recognize objects.

2. Visual localization of a recognized object

3. Depth estimation of geometric distances

4. Recognition of changes of objects in the environment

5. Specific requirements of people with blindness by detecting obstacles and searching for objects and transmitting this information to this target group.

During the project, the following articles on the different topics have been published so far with the participation of members of the AccessibleMaps project:

Robust and Holistic Semantic Scene Understanding

In this research line, we design robust and holistic semantic scene understanding algorithms with the aim to efficiently, reliably, and comprehensively extract and recognize dense objects and stuff information from real-world scenes by using multi-modal data captured from different sensors.

Transfer beyond the Field of View: Dense Panoramic Semantic Segmentation via Unsupervised Domain Adaptation. Jiaming Zhang, Chaoxiang Ma, Kailun Yang, Alina Roitberg, Kunyu Peng, Rainer Stiefelhagen. IEEE Transactions on Intelligent Transportation Systems, 2021.

Exploring Event-driven Dynamic Context for Accident Scene Segmentation. Jiaming Zhang, Kailun Yang, Rainer Stiefelhagen. IEEE Transactions on Intelligent Transportation Systems, 2021.

Is Context-Aware CNN Ready for the Surroundings? Panoramic Semantic Segmentation in the Wild. Kailun Yang, Xinxin Hu, Rainer Stiefelhagen. IEEE Transactions on Image Processing, 2021.

Polarization-driven Semantic Segmentation via Efficient Attention-bridged Fusion. Kaite Xiang, Kailun Yang, Kaiwei Wang. Optics Express, 2021.

NLFNet: Non-Local Fusion Towards Generalized Multimodal Semantic Segmentation across RGB-Depth, Polarization, and Thermal Images. Ran Yan, Kailun Yang, Kaiwei Wang. In IEEE International Conference on Robotics and Biomimetics (ROBIO), Sanya, China, December 2021.

ISSAFE: Improving Semantic Segmentation in Accidents by Fusing Event-based Data. Jiaming Zhang, Kailun Yang, Rainer Stiefelhagen. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Prague, Czech Republic (Virtual), September 2021.

DensePASS: Dense Panoramic Semantic Segmentation via Unsupervised Domain Adaptation with Attention-Augmented Context Exchange. Chaoxiang Ma, Jiaming Zhang, Kailun Yang, Alina Roitberg, Rainer Stiefelhagen. In IEEE International Conference on Intelligent Transportation Systems (ITSC), Indianapolis, IN, United States (Virtual), September 2021.

Aerial-PASS: Panoramic Annular Scene Segmentation in Drone Videos. Lei Sun, Jia Wang, Kailun Yang, Kaikai Wu, Xiangdong Zhou, Kaiwei Wang, Jian Bai. In European Conference on Mobile Robots (ECMR), Bonn, Germany (Virtual), August 2021.

Panoramic Panoptic Segmentation: Towards Complete Surrounding Understanding via Unsupervised Contrastive Learning. Alexander Jaus, Kailun Yang, Rainer Stiefelhagen. In IEEE Intelligent Vehicles Symposium (IV), Nagoya, Japan (Virtual), July 2021. Best Paper Award.

Capturing Omni-Range Context for Omnidirectional Segmentation. Kailun Yang, Jiaming Zhang, Simon Reiß, Xinxin Hu, Rainer Stiefelhagen. In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, United States (Virtual), June 2021.

Omnisupervised Omnidirectional Semantic Segmentation. Kailun Yang, Xinxin Hu, Yicheng Fang, Kaiwei Wang, Rainer Stiefelhagen. IEEE Transactions on Intelligent Transportation Systems, 2020.

Real-time Fusion Network for RGB-D Semantic Segmentation Incorporating Unexpected Obstacle Detection for Road-driving Images. Lei Sun, Kailun Yang, Xinxin Hu, Weijian Hu, Kaiwei Wang. IEEE Robotics and Automation Letters with IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Las Vegas, NV, United States (Virtual), October 2020.

In Defense of Multi-Source Omni-Supervised Efficient ConvNet for Robust Semantic Segmentation in Heterogeneous Unseen Domains. Kailun Yang, Xinxin Hu, Kaiwei Wang, Rainer Stiefelhagen. In IEEE Intelligent Vehicles Symposium (IV), Las Vegas, NV, United States (Virtual), October 2020.

DS-PASS: Detail-Sensitive Panoramic Annular Semantic Segmentation through SwaftNet for Surrounding Sensing. Kailun Yang, Xinxin Hu, Hao Chen, Kaite Xiang, Kaiwei Wang, Rainer Stiefelhagen. In IEEE Intelligent Vehicles Symposium (IV), Las Vegas, NV, United States (Virtual), October 2020.

Universal Semantic Segmentation for Fisheye Urban Driving Images. Yaozu Ye, Kailun Yang, Kaite Xiang, Juan Wang, Kaiwei Wang. In IEEE International Conference on Systems, Man, and Cybernetics (SMC), Toronto, Canada (Virtual), October 2020.

Visual Localization

In this research line, we develop visual localization, odometry, and mapping methods for navigation- and mobility assistance, which can help determine the location in an unknown scene and associate the detected information with the map.

Panoramic annular SLAM with loop closure and global optimization. Hao Chen, Weijian Hu, Kailun Yang, Jian Bai, Kaiwei Wang. Applied Optics, 2021.

Semantic Visual Odometry based on Panoramic Annular Imaging. Hao Chen, Kailun Yang, Weijian Hu, Jian Bai, Kaiwei Wang. Acta Optica Sinica, 2021.

A Panoramic Localizer Based on Coarse-to-Fine Descriptors for Navigation Assistance. Yicheng Fang, Kailun Yang, Ruiqi Cheng, Lei Sun, Kaiwei Wang. Sensors, 2020.

CFVL: A Coarse-to-Fine Vehicle Localizer with Omnidirectional Perception across Severe Appearance Variations. Yicheng Fang, Kaiwei Wang, Ruiqi Cheng, Kailun Yang. In IEEE Intelligent Vehicles Symposium (IV), Las Vegas, NV, United States (Virtual), October 2020.

Depth Estimation for Geometric Understanding

In this research line, we design panoramic, efficient, and robust depth estimation algorithms with the aim of obtaining complete 3D information of real-world scenes, which can facilitate geometric understanding to support mobility assistance.

Panoramic Depth Estimation via Supervised and Unsupervised Learning in Indoor Scenes. Keyang Zhou, Kailun Yang, Kaiwei Wang. Applied Optics, 2021.

PADENet: An Efficient and Robust Panoramic Monocular Depth Estimation Network for Outdoor Scenes. Keyang Zhou, Kaiwei Wang, Kailun Yang. In IEEE Intelligent Transportation Systems Conference (ITSC), Rhodes, Greece (Virtual), September 2020.

A Robust Monocular Depth Estimation Framework Based on Light-Weight ERF-PSPNet for Day-Night Driving Scenes. Keyang Zhou, Kaiwei Wang, Kailun Yang. In International Conference on Graphics, Images and Interactive Techniques (CGIIT), Sanya, China (Virtual), February 2020.

Change Detection

In this research line, we design change detection algorithms to help detect the updates and changes in real-world scenes.

DR-TANet: Dynamic Receptive Temporal Attention Network for Street Scene Change Detection. Shuo Chen, Kailun Yang, Rainer Stiefelhagen. In IEEE Intelligent Vehicles Symposium (IV), Nagoya, Japan (Virtual), July 2021.

Avoidance of obstacles and searching for objects for Visually Impaired People

In this research line, we build mobile assistance systems with the aim of helping visually impaired people safely and independently navigate in real-world scenes, explore unknown spaces, avoid obstacles, and search for objects.

Flying Guide Dog: Walkable Path Discovery for the Visually Impaired Utilizing Drones and Transformer-based Semantic Segmentation. Haobin Tan, Chang Chen, Xinyu Luo, Jiaming Zhang, Constantin Seibold, Kailun Yang, Rainer Stiefelhagen. In IEEE International Conference on Robotics and Biomimetics (ROBIO), Sanya, China, December 2021.

HIDA: Towards Holistic Indoor Understanding for the Visually Impaired via Semantic Instance Segmentation with a Wearable Solid-State LiDAR Sensor. Huayao Liu, Ruiping Liu, Kailun Yang, Jiaming Zhang, Kunyu Peng, Rainer Stiefelhagen. In International Workshop on Assistive Computer Vision and Robotics (ACVR) with IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, Canada (Virtual), October 2021.

Trans4Trans: Efficient Transformer for Transparent Object Segmentation to Help Visually Impaired People Navigate in the Real World. Jiaming Zhang, Kailun Yang, Angela Constantinescu, Kunyu Peng, Karin Müller, Rainer. Stiefelhagen. In International Workshop on Assistive Computer Vision and Robotics (ACVR) with IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, Canada (Virtual), October 2021.

Panoptic Lintention Network: Towards Efficient Navigational Perception for the Visually Impaired. Wei Mao, Jiaming Zhang, Kailun Yang, Rainer Stiefelhagen. In IEEE International Conference on Real-time Computing and Robotics (RCAR), Xining, China, July 2021.

Perception Framework through Real-Time Semantic Segmentation and Scene Recognition on a Wearable System for the Visually Impaired. Yingzhi Zhang, Haoye Chen, Kailun Yang, Jiaming Zhang, Rainer Stiefelhagen. In IEEE International Conference on Real-time Computing and Robotics (RCAR), Xining, China, July 2021.

Helping the Blind to Get through COVID-19: Social Distancing Assistant Using Real-Time Semantic Segmentation on RGB-D Video. Manuel Martinez, Kailun Yang, Angela Constantinescu, Rainer Stiefelhagen. Sensors, 2020.

Can We Unify Perception and Localization in Assisted Navigation? An Indoor Semantic Visual Positioning System for Visually Impaired People. Haoye Chen, Yingzhi Zhang, Kailun Yang, Manuel Martinez, Karin Müller, Rainer Stiefelhagen. In International Conference on Computers Helping People with Special Needs (ICCHP), Lecco, Italy (Virtual), September 2020.

Transmission of information to people with visual impairments

In this research line, we investigate sonification- and interaction-oriented methods to help visually impaired people efficiently and comfortably understand the information acquired by the video analysis systems.

Seeing through Events: Real-Time Moving Object Sonification for Visually Impaired People using Event-Based Camera. Zihao Ji, Weijian Hu, Ze Wang, Kailun Yang, Kaiwei Wang. Sensors, 2021.

Affect-DML: Context-Aware One-Shot Recognition of Human Affect using Deep Metric Learning. Kunyu Peng, Alina Roitberg, David Schneider, Marios Koulakis, Kailun Yang, Rainer Stiefelhagen. In IEEE International Conference on Automatic Face and Gesture Recognition (FG), Jodhpur, India (Virtual), December 2021.

Pose2Drone: A Skeleton-Pose-based Framework for Human-Drone Interaction. Zdravko Marinov, Stanka Vasileva, Qing Wang, Constantin Seibold, Jiaming Zhang, Rainer Stiefelhagen. European Signal Processing Conference (EUSIPCO), Dublin, Ireland (Virtual), May 2021.

A Comparative Study in Real-Time Scene Sonification for Visually Impaired People. Weijian Hu, Kaiwei Wang, Kailun Yang, Ruiqi Cheng, Yaozu Ye, Lei Sun, Zhijie Xu. Sensors, 2020.