Related references
Note: Only part of the references are listed.
Review
Engineering, Civil
Yaodong Cui et al.
Summary: The development of autonomous vehicles has been rapid in recent years, yet achieving full autonomy poses challenges due to the complex and dynamic driving environments. The fusion of camera and LiDAR sensors using deep learning is an emerging research theme. Despite the lack of critical reviews on deep-learning-based camera-LiDAR fusion methods, recent research has focused on leveraging image and point cloud data processing for improved environmental perception and object detection.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Mohamed Hallek et al.
Summary: This study introduces a new stereo matching algorithm that utilizes pixel-level difference adjustment, gradient matching, and rank transform, combined with guided filter aggregation of cost metrics and dynamic programming for disparity calculation to improve accuracy and runtime of the disparity map. Mean-shift image segmentation and refinement techniques are also used to enhance accuracy. The algorithm achieves high disparity evaluation speed and ranks third in accuracy and runtime on the Middlebury stereo benchmark.
JOURNAL OF REAL-TIME IMAGE PROCESSING
(2022)
Article
Mathematics
Matthias Valvekens
Summary: This paper proves that the L-2-Betti numbers of a rigid C*-tensor category vanish when there is an almost-normal subcategory with vanishing L-2-Betti numbers, extending a result from [7]. The criterion is applied to show that the categories constructed from totally disconnected groups in [6] have vanishing L-2-Betti numbers. Additionally, the paper relates the cohomology theory of the quasi-regular inclusion P x Lambda subset of P x Gamma to that of the Schlichting completion G of Lambda < Gamma in the case of an almost-normal inclusion of discrete groups Lambda < Gamma, with Gamma acting on a type II1 factor P by outer automorphisms.
INTERNATIONAL MATHEMATICS RESEARCH NOTICES
(2022)
Article
Computer Science, Artificial Intelligence
James R. Clough et al.
Summary: This method introduces prior knowledge about the topology of segmented objects into the training process of neural networks for image or volume segmentation. The use of persistent homology allows for specifying desired topological features and driving the proposed segmentations to contain these features. The experiments demonstrate the effectiveness of embedding explicit prior knowledge in challenging segmentation tasks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Engineering, Civil
Weixuan Ma et al.
Summary: This paper proposes an algorithm that uses depth visual information to accurately segment roads and vehicles. The method utilizes unsupervised deep learning-based monocular depth estimation and a non-parametric, refined U-V disparity mapping method to obtain the road region of interest, and then uses multi-feature fusion to accurately segment the target region. Experimental results show that the proposed method can accurately and efficiently detect roads and vehicles in a variety of scenarios.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Automation & Control Systems
W. Wang et al.
Summary: An edge detection based method is studied for accurately detecting lane lines in road traffic images at raining weather, which includes algorithms for image enhancement, lane line strengthening, feature point extraction, noise removal, segment connection, and gap filling. Testing on hundreds of images shows satisfactory results compared to traditional algorithms.
INFORMATION TECHNOLOGY AND CONTROL
(2021)
Article
Engineering, Civil
Sampo Kuutti et al.
Summary: Deep learning methods have shown great promise in providing excellent performance for complex and non-linear control problems, as well as generalising previously learned rules to new scenarios. While there have been important advancements in using deep learning for vehicle control, there are still challenges to overcome, such as computation, architecture selection, goal specification, generalisation, verification and validation, as well as safety.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2021)
Article
Engineering, Civil
Di Feng et al.
Summary: Deep learning is driving recent advancements in perception for autonomous driving through the fusion of multiple sensors, but questions regarding network architecture design, fusion timing, and methods remain open. This review aims to systematically summarize methodologies for deep multi-modal object detection and semantic segmentation in autonomous driving, while also discussing challenges and open questions. The reviewed study provides an overview of the topic, fusion methodologies, and offers an interactive online platform for further exploration.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2021)
Article
Automation & Control Systems
Xun Li et al.
Summary: This paper proposes a method of vehicle multi-object identification and classification based on the YOLOv2 algorithm, effectively solving classical multi-object classification problems. The improved algorithm achieved high accuracy and mAP values in both simple and complex backgrounds.
INFORMATION TECHNOLOGY AND CONTROL
(2021)
Article
Astronomy & Astrophysics
Sambit K. Giri et al.
Summary: The study proposes measuring the 3D topology of ionized regions during the epoch of reionization using Betti numbers, which show characteristic evolution and can be fitted with simple analytical functions. The evolution of the Betti numbers is connected with the percolation of ionized and neutral regions, differing between reionization scenarios. Extracting topological information from mock 21-cm observations can constrain reionization models.
MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY
(2021)
Article
Engineering, Mechanical
Guofa Li et al.
Summary: Understanding drivers' behavioral characteristics is crucial for the design of decision-making modules in AVs and ADASs. This paper proposed an unsupervised framework for automatic descriptive driving pattern extraction, which effectively segments driving sequences and clusters them into multiple descriptive patterns. Experimental results demonstrate the effectiveness of the proposed framework in deepening the understanding of drivers' behavioral characteristics.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2021)
Article
Engineering, Civil
Teng Cao et al.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2015)
Article
Computer Science, Artificial Intelligence
Yifei Wang et al.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2014)
Article
Robotics
A. Geiger et al.
INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH
(2013)
Article
Optics
Young-Chul Lim et al.
OPTICAL ENGINEERING
(2011)
Article
Computer Science, Artificial Intelligence
Heiko Hirschmueller
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2008)
