{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,4]],"date-time":"2025-12-04T16:25:20Z","timestamp":1764865520639,"version":"3.46.0"},"reference-count":47,"publisher":"MDPI AG","issue":"12","license":[{"start":{"date-parts":[[2025,12,4]],"date-time":"2025-12-04T00:00:00Z","timestamp":1764806400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"European Space Agency (ESA)\u2019s FutureEO program","award":["4000142334\/23\/I-DT"],"award-info":[{"award-number":["4000142334\/23\/I-DT"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IJGI"],"abstract":"<jats:p>This study explores the impact of synthetic data, both physically based and generatively created, on deep learning analytics for earth observation (EO), focusing on the detection of photovoltaic panels. A YOLOv8 object detection model was trained using a publicly available, multi-resolution very high resolution (VHR) EO dataset (0.8 m, 0.3 m, and 0.1 m), comprising 3716 images from various locations in Jiangsu Province, China. Three benchmarks were established using only real EO data. Subsequent experiments evaluated how the inclusion of synthetic data, in varying types and quantities, influenced the model\u2019s ability to detect photovoltaic panels in VHR imagery. Physically based synthetic images were generated using the Unity engine, which allowed the generation of a wide range of realistic scenes by varying scene parameters automatically. This approach produced not only realistic RGB images but also semantic segmentation maps and pixel-accurate masks identifying photovoltaic panel locations. Generative synthetic data were created using diffusion-based models (DALL\u00b7E 3 and Stable Diffusion XL), guided by prompts to simulate satellite-like imagery containing solar panels. All synthetic images were manually reviewed, and corresponding annotations were ensured to be consistent with the real dataset. Integrating synthetic with real data generally improved model performance, with the best results achieved when both data types were combined. Performance gains were dependent on data distribution and volume, with the most significant improvements observed when synthetic data were used to meet the YOLOv8-recommended minimum of 1500 images per class. In this setting, combining real data with both physically based and generative synthetic data yielded improvements of 1.7% in precision, 3.9% in recall, 2.3% in mAP@50, and 3.3% in mAP@95 compared to training with real data alone. The study also emphasizes the importance of carefully managing the inclusion of synthetic data in training and validation phases to avoid overfitting to synthetic features, with the goal of enhancing generalization to real-world data. Additionally, a pre-training experiment using only synthetic data, followed by fine-tuning with real images, demonstrated improved early-stage training performance, particularly during the first five epochs, highlighting potential benefits in computationally constrained environments.<\/jats:p>","DOI":"10.3390\/ijgi14120481","type":"journal-article","created":{"date-parts":[[2025,12,4]],"date-time":"2025-12-04T16:07:38Z","timestamp":1764864458000},"page":"481","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Impact of Synthetic Data on Deep Learning Models for Earth Observation: Photovoltaic Panel Detection Case Study"],"prefix":"10.3390","volume":"14","author":[{"given":"Enes","family":"Hisam","sequence":"first","affiliation":[{"name":"GMV, Harwell Science and Innovation Campus, Oxford OX11 0RL, Oxfordshire, UK"},{"name":"Geomatics Engineering Program, Graduate School, Istanbul Technical University, 34469 Istanbul, Turkey"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5123-7580","authenticated-orcid":false,"given":"Jesus","family":"Gimeno","sequence":"additional","affiliation":[{"name":"IRTIC, University of Valencia, 46980 Valencia, Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1648-5308","authenticated-orcid":false,"given":"David","family":"Miraut","sequence":"additional","affiliation":[{"name":"GMV, Harwell Science and Innovation Campus, Oxford OX11 0RL, Oxfordshire, UK"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5546-6583","authenticated-orcid":false,"given":"Manolo","family":"P\u00e9rez-Aixendri","sequence":"additional","affiliation":[{"name":"IRTIC, University of Valencia, 46980 Valencia, Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0307-0392","authenticated-orcid":false,"given":"Marcos","family":"Fern\u00e1ndez","sequence":"additional","affiliation":[{"name":"IRTIC, University of Valencia, 46980 Valencia, Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-6345-0111","authenticated-orcid":false,"given":"Rossana","family":"Gini","sequence":"additional","affiliation":[{"name":"GMV, Harwell Science and Innovation Campus, Oxford OX11 0RL, Oxfordshire, UK"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ra\u00fal","family":"Rodr\u00edguez","sequence":"additional","affiliation":[{"name":"GMV, Harwell Science and Innovation Campus, Oxford OX11 0RL, Oxfordshire, UK"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9311-6392","authenticated-orcid":false,"given":"Gabriele","family":"Meoni","sequence":"additional","affiliation":[{"name":"\u03a6-lab, ESRIN, European Space Agency (ESA), 00044 Frascati, Italy"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7498-1540","authenticated-orcid":false,"given":"Dursun Zafer","family":"Seker","sequence":"additional","affiliation":[{"name":"Geomatics Engineering Department, Faculty of Civil Engineering, Istanbul Technical University, 34469 Istanbul, Turkey"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2025,12,4]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"23","DOI":"10.1051\/epjpv\/2023016","article-title":"Snapshot of Photovoltaics\u2212May 2023","volume":"14","year":"2023","journal-title":"EPJ Photovolt."},{"key":"ref_2","unstructured":"Gao, W. (2023). Impact of Renewable Energy Policies on Solar Photovoltaic Energy: Comparison of China, Germany, Japan, and the United States of America. Distributed Energy Resources: Solutions for a Low Carbon Society, Springer International Publishing."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"3474","DOI":"10.1016\/j.egyr.2023.10.022","article-title":"Solar Energy Status in the World: A Comprehensive Review","volume":"10","author":"Pourasl","year":"2023","journal-title":"Energy Rep."},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"He, K., and Zhang, L. (November, January 30). Automatic Detection and Mapping of Solar Photovoltaic Arrays with Deep Convolutional Neural Networks in High Resolution Satellite Images. Proceedings of the 2020 IEEE 4th Conference on Energy Internet and Energy System Integration (EI2), Wuhan, China.","DOI":"10.1109\/EI250167.2020.9347211"},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"636","DOI":"10.1038\/s41597-023-02539-8","article-title":"A Solar Panel Dataset of Very High Resolution Satellite Imagery to Support the Sustainable Development Goals","volume":"10","author":"Clark","year":"2023","journal-title":"Sci. Data"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"59","DOI":"10.1038\/s41597-023-01951-4","article-title":"A Crowdsourced Dataset of Aerial Images with Annotated Solar Photovoltaic Arrays and Installation Metadata","volume":"10","author":"Kasmi","year":"2023","journal-title":"Sci. Data"},{"key":"ref_7","unstructured":"Villalobos, P., Ho, A., Sevilla, J., Besiroglu, T., Heim, L., and Hobbhahn, M. (2024). Will We Run out of Data? Limits of LLM Scaling Based on Human-Generated Data. arXiv."},{"key":"ref_8","unstructured":"Azizi, S., Kornblith, S., Saharia, C., Norouzi, M., and Fleet, D.J. (2023). Synthetic Data from Diffusion Models Improves ImageNet Classification. arXiv."},{"key":"ref_9","unstructured":"Burg, M.F., Wenzel, F., Zietlow, D., Horn, M., Makansi, O., Locatello, F., and Russell, C. (2023). Image Retrieval Outperforms Diffusion Models on Data Augmentation. arXiv."},{"key":"ref_10","unstructured":"Luzi, L., Mayer, P.M., Casco-Rodriguez, J., Siahkoohi, A., and Baraniuk, R.G. (2024). Boomerang: Local Sampling on Image Manifolds Using Diffusion Models. arXiv."},{"key":"ref_11","unstructured":"Veselovsky, V., Ribeiro, M.H., and West, R. (2023). Artificial Artificial Artificial Intelligence: Crowd Workers Widely Use Large Language Models for Text Production Tasks. arXiv."},{"key":"ref_12","unstructured":"Bansal, H., and Grover, A. (2023). Leaving Reality to Imagination: Robust Classification via Generated Datasets. arXiv."},{"key":"ref_13","unstructured":"Dai, H., Liu, Z., Liao, W., Huang, X., Cao, Y., Wu, Z., Zhao, L., Xu, S., Liu, W., and Liu, N. (2023). AugGPT: Leveraging ChatGPT for Text Data Augmentation. arXiv."},{"key":"ref_14","unstructured":"He, R., Sun, S., Yu, X., Xue, C., Zhang, W., Torr, P., Bai, S., and Qi, X. (2023). Is Synthetic Data from Generative Models Ready for Image Recognition?. arXiv."},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Lin, S., Wang, K., Zeng, X., and Zhao, R. (2023). Explore the Power of Synthetic Data on Few-Shot Object Detection. arXiv.","DOI":"10.1109\/CVPRW59228.2023.00071"},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Shipard, J., Wiliem, A., Thanh, K.N., Xiang, W., and Fookes, C. (2023). Diversity is Definitely Needed: Improving Model-Agnostic Zero-Shot Classification via Stable Diffusion. arXiv.","DOI":"10.1109\/CVPRW59228.2023.00084"},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Xu, C., Guo, D., Duan, N., and McAuley, J. (2023). Baize: An Open-Source Chat Model with Parameter-Efficient Tuning on Self-Chat Data. arXiv.","DOI":"10.18653\/v1\/2023.emnlp-main.385"},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Klemp, M., R\u00f6sch, K., Wagner, R., Quehl, J., and Lauer, M. (2023). LDFA: Latent Diffusion Face Anonymization for Self-Driving Applications. arXiv.","DOI":"10.1109\/CVPRW59228.2023.00322"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Packh\u00e4user, K., Folle, L., Thamm, F., and Maier, A. (2023, January 17\u201321). Generation of Anonymous Chest Radiographs Using Latent Diffusion Models for Training Thoracic Abnormality Classification Systems. Proceedings of the 2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI), Cartagena de Indias, Colombia.","DOI":"10.1109\/ISBI53787.2023.10230346"},{"key":"ref_20","unstructured":"Veeravasarapu, V.S.R., Rothkopf, C., and Ramesh, V. (2016). Model-Driven Simulations for Deep Convolutional Neural Networks. arXiv."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Veeravasarapu, V., Rothkopf, C., and Visvanathan, R. (2017, January 27\u201329). Model-Driven Simulations for Computer Vision. Proceedings of the 2017 IEEE Winter Conference on Applications of Computer Vision (WACV), Santa Rosa, CA, USA.","DOI":"10.1109\/WACV.2017.123"},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Savva, M., Kadian, A., Maksymets, O., Zhao, Y., Wijmans, E., Jain, B., Straub, J., Liu, J., Koltun, V., and Malik, J. (November, January 27). Habitat: A Platform for Embodied AI Research. Proceedings of the 2019 IEEE\/CVF International Conference on Computer Vision (ICCV), Seoul, Republic of Korea.","DOI":"10.1109\/ICCV.2019.00943"},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Schwarz, M., and Behnke, S. (August, January 31). Stillleben: Realistic Scene Synthesis for Deep Learning in Robotics. Proceedings of the 2020 IEEE International Conference on Robotics and Automation (ICRA), Paris, France.","DOI":"10.1109\/ICRA40945.2020.9197309"},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"Zhang, Y., Song, S., Yumer, E., Savva, M., Lee, J.-Y., Jin, H., and Funkhouser, T. (2017). Physically-Based Rendering for Indoor Scene Understanding Using Convolutional Neural Networks. arXiv.","DOI":"10.1109\/CVPR.2017.537"},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"e15010","DOI":"10.1111\/cgf.15010","article-title":"Physically Based Real-Time Rendering of Atmospheres Using Mie Theory","volume":"43","author":"Schneegans","year":"2024","journal-title":"Comput. Graph. Forum"},{"key":"ref_26","first-page":"1","article-title":"Synthetic Data Augmentation Using Multiscale Attention CycleGAN for Aircraft Detection in Remote Sensing Images","volume":"19","author":"Liu","year":"2022","journal-title":"IEEE Geosci. Remote Sens. Lett."},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Howe, J., Pula, K., and Reite, A.A. (2019, January 6). Conditional Generative Adversarial Networks for Data Augmentation and Adaptation in Remotely Sensed Imagery. Proceedings of the Applications of Machine Learning, San Diego, CA, USA.","DOI":"10.1117\/12.2529586"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"117","DOI":"10.1145\/3635309","article-title":"Synthetic Data and Hierarchical Object Detection in Overhead Imagery","volume":"20","author":"Clement","year":"2024","journal-title":"ACM Trans Multimed. Comput. Commun. Appl."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"1111","DOI":"10.1109\/LGRS.2017.2699196","article-title":"Synthetic Aperture Radar Image Synthesis by Using Generative Adversarial Nets","volume":"14","author":"Guo","year":"2017","journal-title":"IEEE Geosci. Remote Sens. Lett."},{"key":"ref_30","doi-asserted-by":"crossref","unstructured":"Abady, L., Barni, M., Garzelli, A., and Tondi, B. (2020, January 20). GAN Generation of Synthetic Multispectral Satellite Images. Proceedings of the Image and Signal Processing for Remote Sensing XXVI, SPIE, Online.","DOI":"10.1117\/12.2575765"},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"5389","DOI":"10.5194\/essd-13-5389-2021","article-title":"Multi-Resolution Dataset for Photovoltaic Panel Segmentation from Satellite and Aerial Imagery","volume":"13","author":"Jiang","year":"2021","journal-title":"Earth Syst. Sci. Data"},{"key":"ref_32","doi-asserted-by":"crossref","unstructured":"Yang, Y.-L., Wang, J., Vouga, E., and Wonka, P. (2013). Urban Pattern: Layout Design by Hierarchical Domain Splitting. ACM Trans. Graph., 32.","DOI":"10.1145\/2508363.2508405"},{"key":"ref_33","unstructured":"Gimeno Sancho, J., Fernandez Mar\u00edn, M., Garc\u00eda Blaya, A., P\u00e9rez Folgado, I., Casanova-Salas, P., Gini, R., Fern\u00e1ndez Guirao, A., and P\u00e9rez Aixendri, M. (2024). SD4EO\u2014Physically Based Rendering Images of Human Settlements for Solar Panel Detection. Zenodo."},{"key":"ref_34","unstructured":"Ho, J., Jain, A., and Abbeel, P. (2020). Denoising Diffusion Probabilistic Models. arXiv."},{"key":"ref_35","doi-asserted-by":"crossref","unstructured":"Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. (2021). High-Resolution Image Synthesis with Latent Diffusion Models. arXiv.","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref_36","doi-asserted-by":"crossref","unstructured":"Zhang, L., Rao, A., and Agrawala, M. (2023). Adding Conditional Control to Text-to-Image Diffusion Models. arXiv.","DOI":"10.1109\/ICCV51070.2023.00355"},{"key":"ref_37","doi-asserted-by":"crossref","unstructured":"(2025, November 12). Data Augmentation in Earth Observation: A Diffusion Model Approach. Available online: https:\/\/www.mdpi.com\/2078-2489\/16\/2\/81.","DOI":"10.3390\/info16020081"},{"key":"ref_38","unstructured":"Dhariwal, P., and Nichol, A. (2021). Diffusion Models Beat GANs on Image Synthesis. arXiv."},{"key":"ref_39","doi-asserted-by":"crossref","unstructured":"Fang, H., Han, B., Zhang, S., Zhou, S., Hu, C., and Ye, W.-M. (2024, January 4\u20138). Data Augmentation for Object Detection via Controllable Diffusion Models. Proceedings of the 2024 IEEE\/CVF Winter Conference on Applications of Computer Vision (WACV), Waikoloa, HI, USA.","DOI":"10.1109\/WACV57701.2024.00129"},{"key":"ref_40","doi-asserted-by":"crossref","unstructured":"Zhu, J.-Y., Park, T., Isola, P., and Efros, A.A. (2017). Unpaired Image-To-Image Translation Using Cycle-Consistent Adversarial Networks. arXiv.","DOI":"10.1109\/ICCV.2017.244"},{"key":"ref_41","doi-asserted-by":"crossref","first-page":"1680","DOI":"10.3390\/make5040083","article-title":"A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS","volume":"5","author":"Terven","year":"2023","journal-title":"Mach. Learn. Knowl. Extr."},{"key":"ref_42","doi-asserted-by":"crossref","first-page":"576","DOI":"10.3390\/ai5020030","article-title":"Investigating Training Datasets of Real and Synthetic Images for Outdoor Swimmer Localisation with YOLO","volume":"5","author":"Schneidereit","year":"2024","journal-title":"AI"},{"key":"ref_43","doi-asserted-by":"crossref","first-page":"103809","DOI":"10.1016\/j.compind.2022.103809","article-title":"A Systemic Comparison between Using Augmented Data and Synthetic Data as Means of Enhancing Wafermap Defect Classification","volume":"145","author":"Alqudah","year":"2023","journal-title":"Comput. Ind."},{"key":"ref_44","unstructured":"Liu, R., Wei, J., Liu, F., Si, C., Zhang, Y., Rao, J., Zheng, S., Peng, D., Yang, D., and Zhou, D. (2024). Best Practices and Lessons Learned on Synthetic Data. arXiv."},{"key":"ref_45","unstructured":"Bojar, O., Chatterjee, R., Federmann, C., Fishel, M., Graham, Y., Haddow, B., Huck, M., and Yepes, A.J. (2019). Volume 3: Shared Task Papers, Day 2. Neural Machine Translation of Low-Resource and Similar Languages with Backtranslation. Proceedings of the Fourth Conference on Machine Translation, Florence, Italy, 1\u20132 August 2019, Association for Computational Linguistics."},{"key":"ref_46","doi-asserted-by":"crossref","unstructured":"Sagmeister, D., Sch\u00f6rkhuber, D., Nezveda, M., Stiedl, F., Schimkowitsch, M., and Gelautz, M. (2023, January 4\u20137). Transfer Learning for Driver Pose Estimation from Synthetic Data. Proceedings of the 2023 IEEE Intelligent Vehicles Symposium (IV), Anchorage, AK, USA.","DOI":"10.1109\/IV55152.2023.10186532"},{"key":"ref_47","doi-asserted-by":"crossref","unstructured":"Dolev, S., Katz, J., and Meisels, A. (2022). Transfer Learning for Time Series Classification Using Synthetic Data Generation. Proceedings of the Cyber Security, Cryptology, and Machine Learning, Be\u2019er Sheva, Israel, 19\u201320 December 2024, Springer International Publishing.","DOI":"10.1007\/978-3-031-07689-3"}],"container-title":["ISPRS International Journal of Geo-Information"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2220-9964\/14\/12\/481\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,4]],"date-time":"2025-12-04T16:23:23Z","timestamp":1764865403000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2220-9964\/14\/12\/481"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,4]]},"references-count":47,"journal-issue":{"issue":"12","published-online":{"date-parts":[[2025,12]]}},"alternative-id":["ijgi14120481"],"URL":"https:\/\/doi.org\/10.3390\/ijgi14120481","relation":{},"ISSN":["2220-9964"],"issn-type":[{"value":"2220-9964","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,4]]}}}