Figure - available from: Scientific Data
This content is subject to copyright. Terms and conditions apply.
Geo-located map of drone flyover regions (left, WGS 84 coordinate system, source: Google Maps), DJI M600 drone (upper right), and Zenmuse XT2 camera with a FLIR Tau 2 thermal sensor (lower right). Dashed lines show the flight paths of the drone, polygons the photographed regions. Numbers correspond to identifier of each flight paths, e.g. 2 for Flug1_102 (see Data Records section below). Image source for the drone and camera: © DJI.
Source publication
Thermal Bridges on Building Rooftops (TBBR) is a multi-channel remote sensing dataset. It was recorded during six separate UAV fly-overs of the city center of Karlsruhe, Germany, and comprises a total of 926 high-resolution images with 6927 manually-provided thermal bridge annotations. Each image provides five channels: three color, one thermograph...
Similar publications
The demand for detecting and classifying unmanned aerial vehicle (UAV) objects like birds, planes, and drones is increasing in various fields such as the military, surveillance, etc. A distant object appears to be point size; hence, it is difficult to classify the far-way objects in an image. There is always a trade-off between correct object detec...
Citations
... Condensation of water vapor in building partitions can lead to mold and other fungal developments within the building [1,10,[13][14][15][16]. This phenomenon adversely affects the health and comfort of occupants, leading to serious respiratory, rheumatic, and allergic conditions [17,18]. ...
The junction between the roof and the external wall is a sensitive area within the building envelope; here, increased heat flow often takes place. In the case of partitions insulated with materials based on plant ingredients, thermal bridges are particularly dangerous due to the possibility of condensation and, consequently, mold. The present article analyzed the connection of the roof with the knee wall made of a hemp–lime composite and the ridge in terms of the occurrence of thermal bridges. The following factors that may affect heat transfer in the junction were taken into account: the location of the load-bearing wooden frame, the roof slope, and the presence of internal plaster in the junction. Two-dimensional heat transfer analysis was performed based on the finite element method using THERM 7.4 software. All of the studied thermal bridges had ψ values below 0.10 W/(m·K). Calculations of heat losses through a roof with different slopes were also presented, taking into account the considered thermal bridges. As the roof slope decreases, the heat flow through the roof decreases, despite the increasing value of the linear thermal transmittance. The share of the considered thermal bridges in the total heat loss from the roof reached up to 15%. To verify the obtained results, in further analysis, it would be necessary to calculate the impact of the roof–knee wall bridge variants on heat losses throughout the entire building.
Thermographic inspection is particularly effective in identifying thermal bridges because it visualizes temperature differences on the building’s surface. The focus of this work is on energy-efficient computing for deep learning-based thermal bridge (anomaly) detection models. In this study, we concentrate on object detection-based models such as Mask R-CNN_FPN_50, Swin-T Transformer, and FSAF. We do benchmark tests on TBRR dataset with varying input sizes. To overcome the energy-efficient design, we apply optimizations such as compression, latency reduction, and pruning to these models. After our proposed improvements, the inference of the anomaly detection model, Mask R-CNN_FPN_50 with compression technique, is approximately 7.5% faster than the original. Also, more acceleration is observed in all models with increasing input size. Another criterion we focus on is total energy gain for optimized models. Swin-T transformer has the most inference energy gains for all input sizes (≈\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\approx$$\end{document}27 J for 3000 x 4000 and ≈\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\approx$$\end{document}14 J for 2400 x 3400). In conclusion, our study presents an optimization of size, weight, and power for vision-based anomaly detection for buildings.
