Infrastructure Engineering - Research Publications
Permanent URI for this collection
Search Results
1 - 8 of 8
-
ItemNo Preview AvailableResults of the ISPRS benchmark on indoor modelling(Elsevier BV, 2021-12-01)
-
ItemNo Preview AvailablePose-aware monocular localization of occluded pedestrians in 3D scene space(Elsevier BV, 2021-12-01)
-
ItemParking Occupancy Detection and Slot Delineation Using Deep Learning: A Tutorial(TU Wien Academic Press, 2021)This chapter describes a simple method for parking occupancy detection and an automatic parking slot delineation method using CCTV images. These methods will be presented in the form of MATLAB tutorials with code snippets to allow the interested reader to implement the method and obtain results on a sample dataset. The first tutorial will involve fine-tuning a pre-trained deep neural network for vehicle detection in a sequence of CCTV camera images to determine the occupancy of the parking spaces. In the second tutorial, we perform spatio-temporal analysis of the detections made by a state-of-the-art deep learning object detector (Faster-RCNN) for automatic parking slot delineation.
-
ItemComputer Vision Techniques for Urban Mobility(TU Wien Academic Press, 2021)This chapter provides an overview of computer vision techniques with applications in urban mobility and transport systems. Focusing on imagery and Light Detection and Ranging (LiDAR) point clouds as the main data modalities, the chapter reviews relevant computer vision tasks, including classification, segmentation, object detection and tracking. Example applications of these techniques to data captured by stationary sensors installed in the environment as well as mobile sensors onboard vehicles will then be discussed.
-
ItemSensors for Parking Occupancy Detection(TU Wien Academic Press, 2021)This chapter provides an overview of sensor technologies and methodologies for determining the occupancy of parking spaces. It covers a range of sensors including active and passive sensors that can be installed overhead, in or on the ground in both indoor and outdoor environments. The chapter also provides a comparison of sensors, and Discusses considerations for sensor selection and open challenges in parking occupancy detection.
-
ItemAUGMENTED REALITY ASSET TRACKING USING HOLOLENS(Copernicus GmbH, 2021-06-17)Abstract. Asset Tracking is an essential component of building management process. It involves creating and maintaining a database of detailed information of assets such as location, condition, brand, and type. This information can help building owners make informed decisions for cost-effective maintenance of building assets. Existing approaches to perform asset tracking require a manual process of measuring and recording the asset condition and location, which is labour-intensive and costly. The typical approach usually includes a human operator with pen and paper inspecting the site and manually recording the information about the asset. In this paper, we propose an augmented reality asset tracking system using HoloLens to reduce the manual labour involved in this process. The system can automatically detect the asset, record and update its related information by visual inspection. Assets are detected by feeding images captured by the HoloLens built-in camera to a pre-trained object detection network. Using a combination of various sensor readings from the HoloLens, the system can estimate the location of the asset using visual simultaneous localization and mapping (vSLAM). This information is then viewed and verified by the user using the augmented reality user interface. Upon the user confirmation, this information will be uploaded to a database. As a case study, we demonstrate a vending machine tracking system which is able to detect and localise the vending machines in an indoor environment and create a database of vending machine information. The system can detect vending machines with a mean average precision of 94.8% and a localization accuracy of 2.3 meters without pre-screening or user input.
-
ItemA PLATFORM for MULTILAYERED DOCUMENTATION of CULTURAL HERITAGE(Copernicus GmbH, 2021-06-17)Abstract. This paper presents a platform for multilayered documentation of cultural heritage, inspired by the current lack of a heritage BIM approach capable of creating models with both high geometric accuracy and high semantic richness. The platform is developed in the Unity game engine. It comprises several integrated and interconnected layers or datasets that can include data of different types, such as a point cloud, textured polygonal mesh, parametric information model and images, both 2D images and 360° panoramas. We present an overview of the platform concept, the benefits of the proposed multilayered representation and the details on the implementation and integration of datasets. Also, we present some of the innovative functions made possible by this integration, such as point cloud or mesh cutting and preforming measurements on 2D images and 360° panoramas. We perform and present the results of a preliminary analysis of platform functions, which indicates that the platform can be used for accurate measurement and retrieval of 3D coordinates.
-
ItemGenerating synthetic 3D point segments for improved classification of mobile lidar point clouds(Copernicus GmbH, 2021-06-28)Abstract. Mobile lidar point clouds are commonly used for 3d mapping of road environments as they provide a rich, highly detailed geometric representation of objects on and around the road. However, raw lidar point clouds lack semantic information about the type of objects, which is necessary for various applications. Existing methods for the classification of objects in mobile lidar data, including state of the art deep learning methods, achieve relatively low accuracies, and a primary reason for this under-performance is the inadequacy of available 3d training samples to sufficiently train deep networks. In this paper, we propose a generative model for creating synthetic 3d point segments that can aid in improving the classification performance of mobile lidar point clouds. We train a 3d Adversarial Autoencoder (3dAAE) to generate synthetic point segments that exhibit a high resemblance to and share similar geometric features with real point segments. We evaluate the performance of a PointNet-like classifier trained with and without the synthetic point segments. The evaluation results support our hypothesis that training a classifier with training data augmented with synthetic samples leads to significant improvement in the classification performance. Specifically, our model achieves an F1 score of 0.94 for vehicles and pedestrians and 1.00 for traffic signs.