Sustainable Building Facades Optimization via Deep Learning-Enhanced Segmentation Methods for Improving Indoor Air Quality

IJEP 46(1): 74-88 : Vol. 46 Issue. 1 (January 2026)

Full Text

Shruti Semwal and Garima Verma*

DIT University, Department of Computer Science and Engineering, School of Computing, Dehradun – 248 009, Uttarakhand, India

Abstract

This study seeks to transform the role of green buildings (GB) in sustainable urban development by using deep learning (DL)-based algorithms to segment building facades accurately. By leveraging U-Net models to analyze architectural features, this approach enhances precision in facade segmentation, supporting environmentally conscious design and sustainable urban planning. An open-source dataset of 606 images, annotated with 51,731 architectural components, was used and class imbalances were addressed with six data augmentation techniques. The study started with a base U-Net model (model I) trained on the original dataset, followed by enhanced U-Net model (model II) using canny edge detection (CED) to improve edge clarity. Subsequently, model III was developed by incorporating an attention mechanism in model II. Model evaluations showed that model III achieved the highest performance, delivering detailed facade predictions with an accuracy of 0.99. This research demonstrates that integrating deep learning techniques with edge detection, data augmentation and attention mechanisms significantly improves green buildings segmentation accuracy, offering a valuable tool for architects and urban planners in facade analysis. Enhanced accuracy in architectural feature recognition enables more sustainable urban planning, advancing energy-efficient, low-carbon and eco-friendly construction practices. Ultimately, these methods support urban infrastructure development with a minimal environmental footprint, contributing meaningfully to environmental sustainability.

Keywords

Sustainability, Deep learning, Canny edge detection, Building segmentation, Green buildings, U-Net model

References

  1. Anzagira, L.F., Duah, D. and Badu, E. 2019. A conceptual framework for the uptake of the green building concept in Ghana. Sci. African. 6: e00191. DOI: 10.1016/j.sciaf.2019.e00191.
  2. Ghodrati, N., Samari, M. and Shafiei, M.W.M. 2012. Green buildings impacts on occupants’ health and productivity. J. Appl. Sci. Res., 8(8): 4235-4241. DOI: 10.5555/20133032511.
  3. Santos, T.D.O., Pacheco, F.A.L. and Fernandes, L. F.S. 2024. A systematic analysis on the efficiency and sustainability of green facades and roofs. Sci. Total Env., 932: 173107. DOI: 10.1016/j.scitotenv. 2024.173107.
  4. Allen, J.G., MacNaughton, P., Laurent, J.G.C., Fdanigan, S.S., Eitland, E.S. and Spengler, J.D. 2015. Green buildings and health. Curr. Env. Health Reports. 2(3): 250–258. DOI: 10.1007/s40572-015-0063-y.
  5. Sheweka, S. M. and Arch, N.M.M. 2012. Green facades as a new sustainable approach towards climate change. Energy Procedia. 18: 507-520. DOI: 10.1016/j.egypro.2012.05.062.
  6. Leung, B.C.M. 2018. Greening existing buildings (GEB) strategies. Energy Reports. 4: 159-206. DOI: 10.1016/j.egyr.2018.01.003.
  7. Braulio-Gonzalo, M., Jorge-Ortiz, A. and Bovea, M.D. 2022. How are indicators in green building rating systems addressing sustainability dimensions and life cycle frameworks in residential buildings? Env. Impact Assess. Review. 95: 106793. DOI: 10.1016/j.eiar.2022.106793.
  8. Yin, S., Wu, J., Zhao, J., Nogueira, M. and Lloret, J. 2024. Green buildings: Requirements, features, life cycle and relevant intelligent technologies. In Internet of things and cyber-physical systems. DOI: 10.1016/j.iotcps.2024.09.002.
  9. Guo, Y., Liu, Y., Georgiou, T. and Lew, M.S. 2018. A review of semantic segmentation using deep neural networks. Int. j. multimedia information retrieval. 7: 87-93. DOI: 10.1007/s13735-017-01 41-z.
  10. Zhang, G., Pan, Y. and Zhang, L. 2022. Deep learning for detecting building façade elements from images considering prior knowledge. Automation Construction. 133: 104016. DOI: S09265805210046 72.
  11. El-Hajjar, S., Kassem, H., Abdallah, F. and Omrani, H. 2024. Enhancing building segmentation by deep multiview classification for advancing sustainable urban development. J. Building Eng., 83: 108421.
  12. Cheng, Y., Wang, W., Ren, Z., Zhao, Y., Liao, Y., Ge, Y., Wang, J., He, J., Gu, Y., Wang, Y. and Zhang, W. 2023. Multi-scale feature fusion and transformer network for urban green space segmentation from high-resolution remote sensing images. Int. J. Appl. Earth Observation Geoinf., 124: 103514. DOI: 10.1016/j.jag.2023.103514.
  13. Amirgan, B. and Erener, A. 2024. Semantic segmentation of satellite images with different building types using deep learning methods. Remote Sensing Applications Soc. Env., 34: 101176. DOI: 10.1016/j.rsase.2024.101176.
  14. Almujally, N.A., Chughtai, B.R., Al-Mudawi, N., Alazeb, A., Algarni, A., Alzahrani, H.A. and Park, J. 2024. UNet based on multi-object segmentation and convolution neural network for object recognition. Computers Mater. Continua. 80(1): 1563-1580. DOI: 10.32604/cmc.2024.049333.
  15. Nnolim, U.A. 2020. Automated crack segmentation via saturation channel thresholding, area classification and fusion of modified level set segmentation with Canny edge detection. Heliyon. 6(12): e05748. DOI: 10.1016/j.heliyon.2020.e05748.
  16. Pešek, O., Brodsky, L., Halounova, L., Landa, M. and Boucck, T. 2024. Convolutional neural networks for urban green areas semantic segmentation on Sentinel-2 data. Remote sensing Applications Soc. Env., 36: 101238. DOI: 10.1016/j.rsase.2024.101 238.
  17. Zuo, Y. and Li, W. 2024. An improved UNet lightweight network for semantic segmentation of weed images in corn fields. Computers Mater. Continua. 79(3): 4413-4431. DOI: 10.32604/cmc.2024.04 9805.
  18. Jaradat, H., Alshboul, O.A.M., Obcidat, I.M. and Zoubi, M.K. 2024. Green building, carbon emission and environmental sustainability of construction industry in Jordan: Awareness, actions and barriers. Ain Shams Eng. J., 15(2): 102441. DOI: 10.10 16/j.asej.2023.102441.
  19. Bahr, S. 2024. The relationship between urban greenery, mixed landuse and life satisfaction: An examination using remote sensing data and deep learning. Landscape Urban Planning. 251: 105174. DOI: 10.1016/j.landurbplan.2024.105174.
  20. Liu, X., Deng, Z. and Yang, Y. 2019. Recent progress in semantic image segmentation. Artificial Intelligence Review. 52: 1089-1106.6. DOI: 10.1007/s10462-018-9641-3.
  21. Yang, J., Li, Y., Pan, X. and Zang, J. 2023. Research on building facade feature information extraction. Academic J. Computing Information Sci., 6(7): 72-83. DOI: 10.25236/AJCIS.2023. 060711.
  22. Lu, Y., Wei, W., Li, P., Zhong, T., Nong, Y. and Shi, X. 2023. A deep learning method for building façade parsing utilizing improved Solov2 instance segmentation. Energy Buildings. 295: 113275. DOI: 10.21 39/ssrn.4379439.
  23. Sezen, G., Cakir, M., Atik, M.E. and Duran, Z. 2022. Deep learning-based door and window detection from building façade. Int. Arch. Photogrammetry, Remote Sensing Spatial Information Sci., 43: 315-320. DOI: 10.5194/isprs-archives-XLIII-B4-2022-315-2022.
  24. Dai, M., Ward, W.O.C., Meyers, G., Tingley, D.D. and Mayfield, M. 2021. Residential building facade segmentation in the urban environment. Building Env., 199: 107921.
  25. Rahaman, G.M.A., Längkvist, M. and Loutfi, A. 2024. Deep learning based automated estimation of urban green space index from satellite image: A case study. Urban Forestry Urban Greening. 97: 128373. DOI: 10.1016/j.ufug.2024.128373.
  26. Men, G., He, G. and Wang, G. 2021. Concatenated residual attention UNet for semantic segmentation of urban green space. Forests.12(11): 1441. DOI: 10.3390/f12111441.
  27. Yang, F., Zhang, Y., Jiao, D., Xu, K., Wang, D. and Wang, X. 2023. Detecting window line using an improved stacked hourglass network based on new real-world building façade dataset. Open Geosci., 15(1): 20220476. DOI: 10.1515/geo-2022-0476.
  28. Neuhausen, M., Obel, M., Martin, A., Mark, P. and Konig, M. 2018. Window detection in facade images for risk assessment in tunneling. Visualization Eng., 6: 1. DOI: 10.1186/s40327-018-00 62-9.
  29. Femiani, J., Para, W.R., Mitra, N. and Wonka, P. 2018. Facade segmentation in the wild. arXiv. DOI: 10.48 550/arXiv.1805.08634.
  30. Gu, H. and Choo, S. 2022. Method for constructing a façade dataset through deep learning-based automatic image labelling. Appl. Sci., 12(15): 7570. DOI: 10.3390/app12157570.
  31. Verma, G. 2024. Leveraging smart image processing techniques for early detection of foot ulcers using a deep learning network. Polish J. Radiol., 89: e368-e377. DOI: 10.5114/pjr/189412.
  32. Manilow, E., Seetharaman, P. and Salamon, J. TF representations and masking. In Open source tools and data for music source separation.
  33. Guo, M.H., Xu, T.X., Liu, J.J., Liu, Z.N., Jiang, P.T., Mu, T.J., Zhang, S.H., Martin, R.R., Cheng, M.M. and Hu, S.M. 2022. Attention mechanisms in computer vision: A survey. Computational visual media. 8(3): 331-368. DOI: 10.1007/S41095-022-02 71-Y.