近年來，臺灣政府積極推動能源轉型，由於臺灣具備豐富的太陽能資源，且整體呈現高建物密集與高都市化的土地使用型態，因此政府將屋頂型太陽光電視為重點發展的再生能源，而屋頂太陽光電潛力評估對於促進屋頂型太陽光電的部署至關重要，潛力評估結果可使相關人士得以分析適合開發太陽光電的建築物與地區，從而最大限度地利用太陽能資源。
在屋頂太陽光電可用面積推估方法中，基於GIS建模的方法被認為是較實用且有效的技術，然而該方法經常需以較耗時的手動檢查方式幫助分析屋頂物體對可用屋頂面積的影響，或使用空間分辨率≤ 0.25m/pixel的DSM數據才足以對屋頂物體有良好的辨識效果，而要取得如此高精度的DSM數據相對困難。因此本研究嘗試以Mask R-CNN實例分割模型改進基於GIS建模的方法對於屋頂物體辨識的限制，開發出ㄧ基於GIS建模與Mask R-CNN的屋頂太陽光電潛力評估方法，並以南科台南園區做為研究區域，在考量多個屋頂適宜性因子的情況下分析該區域所具備的屋頂太陽光電潛力。
潛力評估結果顯示本研究區域中適合開發屋頂型太陽光電的建築物比例達85.63%，且四種不同建築類型的比例均達80%以上，可用於裝設太陽能板的屋頂面積為866,553.7m2，可裝設約129982.95kW的太陽能板，相當於2025年太陽光電目標裝置容量的0.65%，而每年總計可產生約223.6 GWh的太陽能發電量，相當於2025年太陽光電發展目標預估發電量的0.89%，故本研究區域具備一定的開發價值。Mask R-CNN屋頂物體辨識模型的最佳性能評估指標結果如下：Precision為0.6988、Recall為0.6017、F1-Score為0.6466，整體表現一般，然而在僅考量屋頂物體為適宜性因子的情況下，以Mask R-CNN模型與手動選擇方法所得出的各月技術潛力平均相對百分比差異為1.03%，可推斷使用Mask R-CNN模型分析大範圍區域的屋頂物體對屋頂太陽光電潛力的影響時，具備一定的可靠性，並且能更快地獲得辨識結果以及擴展研究規模。

In recent years, the Taiwanese government has been actively promoting energy transition. Due to Taiwan's abundant solar energy resources, its high concentration of buildings, and urbanized land use patterns, the government has identified rooftop photovoltaic (PV) as a key focus for renewable energy development. Evaluating rooftop PV potential is crucial in promoting the deployment of rooftop PV systems. In previous research, the GIS-based modeling methods have been considered practical and effective for estimating the available area for rooftop PV. However, the GIS-based modeling methods often require manual inspection to assist in analyzing the influence of roof objects on the available roof area. Alternatively, high-resolution Digital Surface Model (DSM) data with a spatial resolution of ≤ 0.25m is necessary to accurately identify roof objects. However, acquiring DSM data with such high precision is a challenging task. Therefore, our study attempts to improve the limitations of rooftop object identification in the GIS-based modeling methods by integrating the Mask R-CNN instance segmentation model. We consider the Tainan Science Park as the research area and aim to analyze the rooftop photovoltaic potential in this region while considering multiple roof suitability factors.

Athanasioss, A. D., Biberacher, M., Dominguez, J., Fiorese, G., Gadocha, S., Gnansounou, E., Guariso, G., Kartalidis, A., Panichelli, L., Pinedo, I. & Robba, M. (2011). Methods and tools to evaluate the availability of renewable energy sources. Renewable and Sustainable Energy Reviews, 15 (2), 1182-1200.
Assouline, D., Mohajeri, N. & Scartezzini, J. L. (2017). Quantifying rooftop photovoltaic solar energy potential: A machine learning approach. Solar Energy. 141, 278-296.
Assouline, D., Mohajeri, N. & Scartezzini, J. L. (2018). Large-scale rooftop solar photovoltaic technical potential estimation using Random Forests. Applied Energy, 217, 189-211.
Attard, L., Debono, C. J., Valentino, G., Castro, M. D., Masi, A. & Scibile, L. (2019). Automatic Crack Detection using Mask R-CNN. 2019 11th International Symposium on Image and Signal Processing and Analysis (ISPA).
Alouta, R. & Hess, K. (ESRI). (2021). Deep Learning with ArcGIS Pro Tips & Tricks: Part 1. https://www.esri.com/arcgis-blog/products/arcgis-pro/imagery/deep-learning-with-arcgis-pro-tips-tricks/
Alouta, R. & Hess, K. (ESRI). (2021). Deep Learning with ArcGIS Pro Tips & Tricks: Part 2. https://www.esri.com/arcgis-blog/products/arcgis-pro/mapping/deep-learning-with-arcgis-pro-tips-tricks-part-2/
Alina, W., Roberto, C., Nahid, M. & Louis, S. J. (2020). A fast machine learning model for large-scale estimation of annual solar irradiation on rooftops. IEA SHC International Conference on Solar Heating and Cooling for Buildings and Industry 2019.
Burrough, P. A. & McDonell, R. A. (1998). Principles of Geographical Information Systems. Oxford University Press, pp. 190.
Byrne, J., Taminiau, J., Kurdgelashvili, L., & Kim, K. N. (2015). A review of the solar city concept and methods to assess rooftop solar electric potential, with an illustrative application to the city of Seoul. Renewable and Sustainable Energy Reviews, 41, 830-844.
Bernstein, L., Bosch, P. & Canziani, O. (2007). AR4 Climate Change 2007. IPCC, 104.
Boateng, M. A., Sey, N. E. N., Amproche, A. A. & Domfeh, M. K. (2022). Instance segmentation scheme for roofs in rural areas based on Mask R-CNN. The Egyptian Journal of Remote Sensing and Space Science, 25 (2), 569-577.
Camargo, L. R., Zink, R., Dorner, W. & Stoeglehner, G. (2015). Spatio-temporal modeling of roof-top photovoltaic panels for improved technical potential assessment and electricity peak load offsetting at the municipal scale. Computers, Environment and Urban Systems, 52, Pages 58-69.
Chow, A., Li, S. & Fung, A. S. (2016). Modeling urban solar energy with high spatiotemporal resolution: A case study in Toronto, Canada. International Journal of Green Energy, 13 (11).
Cook, R. L., Porter, T. & Carpenter, L. (1984). Distributed ray tracing. Computer Graphics, 18 (3).
Castellanos, S., Sunter, D. A. & Kammen, D. M. (2017). Rooftop solar photovoltaic potential in cities: how scalable are assessment approaches? Environmental Research Letters, 12 (12).
Esri：How Slope works. ArcGIS Pro. https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-slope-works.htm
Esri：How Aspect works. ArcGIS Pro. https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-aspect-works.htm
Esri：Modeling solar radiation. ArcGIS Pro. https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/modeling-solar-radiation.htm
Esri：Compute Accuracy For Object Detection. ArcGIS Pro. https://pro.arcgis.com/en/pro-app/latest/tool-reference/image-analyst/compute-accuracy-for-object-detection.htm
Esri：How Compute Accuracy For Object Detection works. ArcGIS Pro. https://pro.arcgis.com/en/pro-app/latest/tool-reference/image-analyst/how-compute-accuracy-for-object-detection-works.htm
Esri：How Hillshade works. ArcGIS Pro. https://pro.arcgis.com/en/pro-app/3.0/tool-reference/spatial-analyst/how-hillshade-works.htm
Esri：Regularize Building Footprint. ArcGIS Pro. https://pro.arcgis.com/en/pro-app/latest/tool-reference/3d-analyst/regularize-building-footprint.htm
Esri：Overview of the ArcGIS API for Python. https://developers.arcgis.com/python/guide/overview-of-the-arcgis-api-for-python/
Eltawil, M. A. & Zhao, Z. (2010). Grid-connected photovoltaic power systems: Technical and potential problems—A review. Renewable and Sustainable Energy Reviews, 14 (1), 112-129.
Falayi, E. & Rabiu, A. B. (2012). Solar Radiation Models and Information for Renewable Energy Applications. Solar Radiation.
Fu, P. & Rich, P. M. (2000). A geometric solar radiation model with applications in agriculture and forestry. Comput Electron Agric, 37, 25-35.
Fathizad, H., Mobin, M. H., Gholamnia, A. & Sodaiezadeh, H. (2017). Modeling and mapping of solar radiation using geostatistical analysis methods in Iran. Arabian Journal of Geosciences, 10 (391).
Gooding, J., Crook, R. & Tomlin, A. S. (2015). Modelling of roof geometries from low-resolution LiDAR data for city-scale solar energy applications using a neighbouring buildings method. Applied Energy, 15, 93-104.
Gergelova, M. B., Labant, S., Kuzevic, S., Kuzevic, Z. & Pavolova, H. (2020). Identification of Roof Surfaces from LiDAR Cloud Points by GIS Tools: A Case Study of Lučenec, Slovakia. Sustainability, 12 (17).
Gergelova, M. B., Kuzevic, S., Labant, S., Kuzevic, Z., Bobikova, D. & Mizak, J. (2020). Roof’s Potential and Suitability for PV Systems Based on LiDAR: A Case Study of Komárno, Slovakia. Sustainability, 12 (23).
Gutschner, M., Nowak, S., Ruoss, D., Togweiler, P. & Schoen, T. (IEA). (2002). Potential for building integrated photovoltaics. Report IEA - PVPS T7-4 : 2002 (Summary).
Gagnon, P., Margolis, R., Melius, J., Phillips, C. & Elmore, R. (NREL). (2016). Rooftop Solar Photovoltaic Technical Potential in the United States: A Detailed Assessment. NREL/TP-6A20-65298.
Gu, W., Bai, S. & Kong, L. (2022). A review on 2D instance segmentation based on deep neural networks. Image and Vision Computing, 120, 104401.
Gassar, A. A. A. & Cha, S. H. (2021). Review of geographic information systems-based rooftop solar photovoltaic potential estimation approaches at urban scales. Applied Energy, 291 (1), 116817.
Goutte, C. & Gaussier, E. (2005). A Probabilistic Interpretation of Precision, Recall and F-Score, with Implication for Evaluation. Advances in Information Retrieval, 345-359.
He, K., Gkioxari, G., Dollár, P. & Girshick, R. (2017). Mask R-CNN. International Conference on Computer Vision (ICCV), 0.1109/ICCV.2017.322.
He, K. & Zhang, L. (2020). Automatic Detection and Mapping of Solar Photovoltaic Arrays with Deep Convolutional Neural Networks in High Resolution Satellite Images. 2020 IEEE 4th Conference on Energy Internet and Energy System Integration (EI2).
He, K., Zhang, X., Ren, S. & Sun, J. (2015). Deep Residual Learning for Image Recognition. Computer Vision and Pattern Recognition (cs.CV).
Hazarika, M. K., Nashrrullah, S., Alam, M., Chen, L. & Westen, C. V. (2016). Building footprint maps. Methodology Book section 5.2: Characterization of assets.
Henderson, P. & Ferrari, V. (2016). End-to-End Training of Object Class Detectors for Mean Average Precision. Computer Vision – ACCV 2016 , 198-213.
Hong, T., Lee, M., Koo, C., Jeong, K.& Kim, J. (2017). Development of a method for estimating the rooftop solar photovoltaic (PV) potential by analyzing the available rooftop area using Hillshade analysis. Applied Energy, 194 (15), 320-332.
Jurasz, J. K., Dąbek, P. B. & Campana, P. E. (2020). Can a city reach energy self-sufficiency by means of rooftop photovoltaics? Case study from Poland. Journal of Cleaner Production, 245 (1), 118813.
Jaroslav, H. (2016). Topographic Solar Radiation Modeling for Environmental Applications. Springer, Solar Energy, 715-730.
Kirillov, A., He, K., Girshick, R., Rother, C., Dollar´, P., Facebook AI Research (FAIR) & Heidelberg University. (2019). Panoptic Segmentation. Computer Vision and Pattern Recognition (cs.CV).
Kodysh, J. B., Omitaomu, O. A., Bhaduri, B. L. & Neish, B. S. (2013). Methodology for estimating solar potential on multiple building rooftops for photovoltaic systems. Sustainable Cities and Society, 8, 31-41.
Köppen, W. & Geiger, R. (1936). Das geographische System der Klimate. Handbuch der Klimatologie. Vol. 1. Berlin: Borntraeger.
Ko, L., Wang, J. C., Chen, C. Y. & Tsai, H. Y. (2015). Evaluation of the development potential of rooftop solar photovoltaic in Taiwan. Renewable Energy, 76, 582-595.
Kouhestani, F. M., Byrne, J., Johnson, D., Spencer, L., Hazendonk, P. & Brown, B. (2019). Evaluating solar energy technical and economic potential on rooftops in an urban setting: the city of Lethbridge, Canada. International Journal of Energy and Environmental Engineering volume 10, pages13–32.
Liang, S. M., Qi, F. Y., Ding, Y. F., Cao, R., Yang, Q. & Yan, W. (2020). Mask R-CNN based segmentation method for satellite imagery of photovoltaics generation systems. 2020 39th Chinese Control Conference (CCC).
Lin, T. Y., Maire, M., Belongie, S., Bourdev, L., Girshick, R., Hays, J., Perona, P., Ramanan, D., Zitnick, C. L. & Dollar, P. (2015). Microsoft COCO: Common Objects in Context. Computer Vision and Pattern Recognition (cs.CV).
Melius, J., Margolis, R. & Ong, S. (NREL). (2013). Estimating Rooftop Suitability for PV: A Review of Methods, Patents, and Validation Techniques. NREL/TP-6A20-60593.
Mohajeri, N., Assouline, D., Guiboud, B., Bill, A., Gudmundsson, A. & Scartezzini, J. L. (2018). A city-scale roof shape classification using machine learning for solar energy applications. Renewable Energy, 121, 81-93.
Moudrý, V., Beková, A, & Lagner, O. (2019). Evaluation of a high resolution UAV imagery model for rooftop solar irradiation estimates. Remote Sensing Letters, 10 (11).
Nelson, J. R. & Grubesic, T. H. (2020). The use of LiDAR versus unmanned aerial systems (UAS) to assess rooftop solar energy potential. Sustainable Cities and Society, 61, 102353.
Ordóñez, J., Jadraque, E., Alegre J. & Martínez G. (2010). Analysis of the photovoltaic solar energy capacity of residential rooftops in Andalusia (Spain). Renewable and Sustainable Energy Reviews, 14 (7), 2122-2130.
Parida, B., Iniyan, S. & Goic, R. (2011). A review of solar photovoltaic technologies. Renewable and Sustainable Energy Reviews, 15 (3), 1625-1636.
Palmer, D., Koumpli, E., Cole, I., Gottschalg, R. & Betts, T. (2018). A GIS-Based Method for Identification of Wide Area Rooftop Suitability for Minimum Size PV Systems Using LiDAR Data and Photogrammetry. Energies , 11 (12), 3506.
Peng, J. & Lu, L. (2013). Investigation on the development potential of rooftop PV system in Hong Kong and its environmental benefits. Renewable and Sustainable Energy Reviews, 27, 149-162.
Rodríguez, J. S. (2013). Photovoltaic techno-economical potential on roofs in regions and islands: The case of the Canary Islands. Methodological review and methodology proposal. Renewable and Sustainable Energy Reviews, 20, 219-239.
Rodrígueza, L. R., Duminil, E., Ramos, J. S. & Eicker, U. (2017). Assessment of the photovoltaic potential at urban level based on 3D city models: A case study and new methodological approach. Solar Energy, 146, 264-275.
Redweik, P., Catita, C. & Brito, M. (2013). Solar energy potential on roofs and facades in an urban landscape. Solar Energy, 97, 332-341.
Ren, S., He, K., Girshick, R. & Sun, J. (2016). Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. Computer Vision and Pattern Recognition (cs.CV).
Suomalainen, K., Wang, V. & Sharp, B. (2017). Rooftop solar potential based on LiDAR data: Bottom-up assessment at neighbourhood level. Renewable Energy, 111, 463-475.
Schulz, M., Boughattas, B. & Wendel, F. (2021). DetEEktor: Mask R-CNN based neural network for energy plant identification on aerial photographs. Energy and AI, 5, 100069.
Singh, R. (2019). How we did it: End-to-end deep learning in ArcGIS. Geospatial Artificial Intelligence.
Singh, R. (ESRI). (2020). Deep learning models in arcgis.learn. https://www.esri.com/arcgis-blog/products/api-python/analytics/deep-learning-models-in-arcgis-learn/
Schuffert, S., Voegtle, T., Tate, N. & Ramirez, A. (2015). Quality Assessment of Roof Planes Extracted from Height Data for Solar Energy Systems by the EAGLE Platform. Remote Sens, 7 (12), 17016-17034.
Stack, V. & Narine, L. L. (2022). Sustainability at Auburn University: Assessing Rooftop Solar Energy Potential for Electricity Generation with Remote Sensing and GIS in a Southern US Campus. Sustainability , 14 (2), 626
Sulaiman, A. (2021). Deep Learning with Python and ArcGIS: Detecting Solar Panels. https://python.plainenglish.io/deep-learning-in-arcgis-pro-detecting-solar-panels-e291edc4e8d1
Thebault, M., Clivillé, V., Berrah, L. & Desthieux, G. (2020). Multicriteria roof sorting for the integration of photovoltaic systems in urban environments. Sustainable Cities and Society, 60, 102259.
U.S. Environmental Protection Agency (USEPA). (2021). Green Power Equivalency Calculator - Calculations and References. https://www.epa.gov/green-power-markets/green-power-equivalency-calculator-calculations-and-references
Verso, A., Martin, A., Amador, J. & Dominguez, J. (2015). GIS-based method to evaluate the photovoltaic potential in the urban environments: The particular case of Miraflores de la Sierra. Solar Energy, 117, 236-245.
Wiginton, L.K., Nguyen, H.T. & Pearce, J.M. (2010). Quantifying rooftop solar photovoltaic potential for regional renewable energy policy. Computers, Environment and Urban Systems, 34 (4), 345-357.
Wegertseder, P., Lund, P., Mikkola, J., Alvarado, R. G. (2016). Combining solar resource mapping and energy system integration methods for realistic valuation of urban solar energy potential. Solar Energy, 135, 325-336.
Yue, C. Y. & Huang, G. R. (2011). An evaluation of domestic solar energy potential in Taiwan incorporating land use analysis. Energy Policy, 39 (12), 7988-8002.
Yuan, J., Farnham, C., Emura, K. & Lu, S. (2016). A method to estimate the potential of rooftop photovoltaic power generation for a region. Urban Climate, 17, 1-19.
Zhao, K., Kang, J., Jung, J. & Sohn, G. (2018). Building Extraction from Satellite Images Using Mask R-CNN with Building Boundary Regularization. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops.
Zink, R., Camargo, L. R., Reidelstürz, P. & Dorner, W. (2015). Photogrammetric Point Clouds for GIS-Based High-Resolution Estimation of Solar Radiation for Roof-Top Solar Systems. Surface Models for Geosciences, 293-305.
Zhang, X., Walker, R., Salisbury, M., Hromiko, R. & Schreiber, J. (2009). Creating a Solar City: Determining the Potential of Solar Rooftop Systems in the City of Newark. Renewable Energy Applications for Delaware Yearly (READY) Project.
陳儀安（2011）。以日照陰影觀點探討都市建物之太陽光電設置潛力-以台南市中西區一街廓為例。國立成功大學都市計劃研究所碩士論文，台南市。
徐天佑、曾鴻陽（2007）。台灣地區有關太陽能日照量之環境時空因素研究探討。環境教育學刊，第六期。
宋昱賢、洪偉諺、薛健宏（2017），陽光屋頂「昕」希望 - 以彰化市中小學熱區分布範圍為例。2017國土測繪圖資GIS專題應用競賽。
許揚典（2020）。運用日照遮蔽及輻射量評估屋頂型太陽能發展潛力-以臺中市為例。臺中市政府109年度市政發展研究論文獎助計畫。
劉家維（2016）。以地理統計方法繪製台灣地區太陽能輻射量分布。國立成功大學航空太空工程學系碩士論文，台南市。
蕭子訓、黃孔良、張耀仁（2015）。太陽光電年發電量預測模型建置及策略研究。台灣能源期刊，第四卷第四期，401-430頁。
黃瓊瑩（2017）。太陽輻照度對太陽能發電影響-於住宅應用之研究。國立中正大學雲端計算與物聯網數位學習碩士在職專班碩士論文。
陶政慷（1995）。三維地理資訊系統在都市建築日照之應用-以陰影遮蔽及太陽輻射效能分析為例。國立中正大學海洋環境及工程研究所碩士論文。
詹肇裕（2005）。太陽光電技術應用於建築設計之評估與策略研究。台北國立台灣科技大學建築系博士論文。
徐翠華（2002）。台灣地區太陽輻射及太陽能發電潛力之研究。國立臺灣師範大學地理研究所碩士論文。
柯酈（2014）。臺灣屋頂型太陽光電開發潛力之研究。國立成功大學資源工程學系博士論文。
工業技術研究院（2022）。家庭用電資訊百科。取自：https://www.energypark.org.tw/save-energy-resource/books/energy-saving-manual.html?task=convert.getpdf&id=84&filename=2021family%20book_2022.7.28update.pdf　
南科建築管理課（2019）。南部科學園區台南園區土地使用計畫示意圖。取自：https://w3.stsp.gov.tw/web/WEB/Jsp/Land/default.jsp?pageID=4550　
台灣地質知識服務網（2023）。數值地形模型Digital Terrain Model（DTM）。取自：https://twgeoref.moeacgs.gov.tw/GipOpenWeb/wSite/ct?ctNode=1233&mp=105&xItem=140861
中央氣象局（2022）。臺灣四季太陽仰角與方位角。取自：https://www.cwb.gov.tw/Data/astronomy/season.pdf　
中央氣象局（2022）。觀測資料查詢。取自：https://e-service.cwb.gov.tw/HistoryDataQuery/　
台灣電力股份有限公司。台灣電力公司_各縣市住宅、服務業及機關用電統計資料。取自：https://data.gov.tw/dataset/29935