Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Deep learning approaches to building rooftop thermal bridge detection from aerial images
https://orcid.org/0000-0002-9082-9095
https://orcid.org/0000-0002-8517-2359
https://orcid.org/0000-0002-9822-244X
https://orcid.org/0000-0002-2233-1041
https://orcid.org/0000-0001-9930-5354
Abstract Thermal bridges are weak points of building envelopes that can lead to energy losses, collection of moisture, and formation of mould in the building fabric. To detect thermal bridges of large building stocks, drones with thermographic cameras can be used. As the manual analysis of comprehensive image datasets is very time-consuming, we investigate deep learning approaches for its automation. For this, we focus on thermal bridges on building rooftops recorded in panorama drone images from our updated dataset of Thermal Bridges on Building Rooftops (TBBRv2), containing 926 images with 6,927 annotations. The images include RGB, thermal, and height information. We compare state-of-the-art models with and without pretraining from five different neural network architectures: MaskRCNN R50, Swin-T transformer, TridentNet, FSAF, and a MaskRCNN R18 baseline. We find promising results, especially for pretrained models, scoring an Average Recall above for detecting large thermal bridges with a pretrained Swin-T Transformer model.
Highlights Updated Thermal Bridges on Building Rooftops (TBBRv2) dataset. Height map data benefits for detection of thermal bridges on rooftops investigated. In-depth study into five SOTA neural networks for thermal bridge detection. All code, configuration files, and the TBBRv2 dataset made publicly available online.
Deep learning approaches to building rooftop thermal bridge detection from aerial images
https://orcid.org/0000-0002-9082-9095
https://orcid.org/0000-0002-8517-2359
https://orcid.org/0000-0002-9822-244X
https://orcid.org/0000-0002-2233-1041
https://orcid.org/0000-0001-9930-5354
Abstract Thermal bridges are weak points of building envelopes that can lead to energy losses, collection of moisture, and formation of mould in the building fabric. To detect thermal bridges of large building stocks, drones with thermographic cameras can be used. As the manual analysis of comprehensive image datasets is very time-consuming, we investigate deep learning approaches for its automation. For this, we focus on thermal bridges on building rooftops recorded in panorama drone images from our updated dataset of Thermal Bridges on Building Rooftops (TBBRv2), containing 926 images with 6,927 annotations. The images include RGB, thermal, and height information. We compare state-of-the-art models with and without pretraining from five different neural network architectures: MaskRCNN R50, Swin-T transformer, TridentNet, FSAF, and a MaskRCNN R18 baseline. We find promising results, especially for pretrained models, scoring an Average Recall above for detecting large thermal bridges with a pretrained Swin-T Transformer model.
Highlights Updated Thermal Bridges on Building Rooftops (TBBRv2) dataset. Height map data benefits for detection of thermal bridges on rooftops investigated. In-depth study into five SOTA neural networks for thermal bridge detection. All code, configuration files, and the TBBRv2 dataset made publicly available online.
Deep learning approaches to building rooftop thermal bridge detection from aerial images
https://orcid.org/0000-0002-9082-9095
https://orcid.org/0000-0002-8517-2359
https://orcid.org/0000-0002-9822-244X
https://orcid.org/0000-0002-2233-1041
https://orcid.org/0000-0001-9930-5354
Abstract Thermal bridges are weak points of building envelopes that can lead to energy losses, collection of moisture, and formation of mould in the building fabric. To detect thermal bridges of large building stocks, drones with thermographic cameras can be used. As the manual analysis of comprehensive image datasets is very time-consuming, we investigate deep learning approaches for its automation. For this, we focus on thermal bridges on building rooftops recorded in panorama drone images from our updated dataset of Thermal Bridges on Building Rooftops (TBBRv2), containing 926 images with 6,927 annotations. The images include RGB, thermal, and height information. We compare state-of-the-art models with and without pretraining from five different neural network architectures: MaskRCNN R50, Swin-T transformer, TridentNet, FSAF, and a MaskRCNN R18 baseline. We find promising results, especially for pretrained models, scoring an Average Recall above for detecting large thermal bridges with a pretrained Swin-T Transformer model.
Highlights Updated Thermal Bridges on Building Rooftops (TBBRv2) dataset. Height map data benefits for detection of thermal bridges on rooftops investigated. In-depth study into five SOTA neural networks for thermal bridge detection. All code, configuration files, and the TBBRv2 dataset made publicly available online.
Deep learning approaches to building rooftop thermal bridge detection from aerial images
Mayer, Zoe (Autor:in) / Kahn, James (Autor:in) / Hou, Yu (Autor:in) / Götz, Markus (Autor:in) / Volk, Rebekka (Autor:in) / Schultmann, Frank (Autor:in)
26.11.2022
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Deep learning approaches to building rooftop thermal bridge detection from aerial images
| DataCite | 2023
AI-based thermal bridge detection of building rooftops on district scale using aerial images
| DataCite | 2021
Deep learning application for urban change detection from aerial images
| BASE | 2021