[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-2163":3},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":6,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":23,"tags":25,"view_count":31,"doi":32,"paper":33,"created_at":49},2163,"A Hybrid Transfer Learning Framework for Seasonal Classification of Satellite Images","https:\u002F\u002Fdoi.org\u002F10.29109\u002Fgujsc.1958797","Seasonal classification from satellite imagery is an important remote sensing task for monitoring vegetation dynamics, agricultural processes, environmental change, and climate-related spatial patterns. However, developing robust deep learning models for this task is challenging due to limited labeled data, regional variability, and the computational cost of training large-scale networks from scratch. This study proposes a hybrid transfer learning-based framework for seasonal classification using satellite images collected from 81 provinces of Türkiye. A custom dataset was constructed from monthly satellite images, and eight pretrained deep learning architectures were evaluated as feature extractors. The extracted deep representations were classified using seven machine learning algorithms. The experimental results showed that both the choice of pretrained feature extractor and the classifier affect seasonal classification performance. Among models, ConvNeXt combined with the Multi-Layer Perceptron achieved the best performance. Based on the comparative analysis, ConvNeXt, Vision Transformer, and Swin Transformer were selected as the top three feature extractors, while the Multi-Layer Perceptron was selected as the final classifier. The proposed framework provides an effective and computationally practical approach for seasonal classification and offers a promising basis for future environmental monitoring and agricultural remote sensing applications.","基于卫星影像的季节分类是一项重要的遥感任务，可用于监测植被动态、农业过程、环境变化以及与气候相关的空间格局。然而，由于标注数据有限、区域差异以及从零开始训练大规模网络的计算成本，开发用于该任务的稳健深度学习模型具有挑战性。本研究提出了一种基于混合迁移学习的框架，利用从土耳其81个省份收集的卫星影像进行季节分类。研究构建了一个由月度卫星影像组成的自定义数据集，并评估了八种预训练深度学习架构作为特征提取器的效果。提取出的深层表示使用七种机器学习算法进行分类。实验结果表明，预训练特征提取器和分类器的选择均会影响季节分类性能。在各类模型中，ConvNeXt结合多层感知机取得了最佳性能。基于对比分析，ConvNeXt、Vision Transformer和Swin Transformer被选为排名前三的特征提取器，而多层感知机被选为最终分类器。所提出的框架为季节分类提供了一种有效且计算上实用的方法，并为未来环境监测和农业遥感应用提供了有前景的基础。",null,"Gazi Üniversitesi Fen Bilimleri Dergisi Part C Tasarım ve Teknoloji","2026-09-10T00:00:00Z","论文",10,false,66,{"impact":17,"substance":18,"depth":19,"authority":17,"freshness":20,"relevant":21,"comment":22},12,18,16,8,1,"基于土耳其81省卫星影像的迁移学习季节分类框架，方法对比扎实、结论可靠，对农业遥感监测有参考价值，但属学术论文且非国内应用，影响力有限。",[24],{"name":10,"url":6},[26,27,28,29,30],"智慧农业","农业人工智能","遥感","作物监测","迁移学习",0,"10.29109\u002Fgujsc.1958797",{"doi":32,"openalex_id":34,"authors":35,"venue":10,"cited_by_count":31,"oa_url":6,"card":42,"direction":46,"ingested_from":48},"W7212161719",[36,39],{"name":37,"orcid":38},"Eyyüp YILDIZ","https:\u002F\u002Forcid.org\u002F0000-0002-7051-3368",{"name":40,"orcid":41},"Özge Aslan Yıldız","https:\u002F\u002Forcid.org\u002F0000-0001-7688-9326",{"tldr":43,"method":44,"finding":45,"direction":46,"opportunity":47},"提出混合迁移学习框架，用预训练模型提取特征并结合机器学习分类器实现卫星图像季节分类。","基于土耳其81省月度卫星图像构建数据集，评估8种预训练模型和7种分类器。","ConvNeXt结合多层感知机表现最佳，特征提取器和分类器选择均影响性能。","农业遥感与作物表型","可探索该框架在作物物候监测、跨区域迁移及多时相农业遥感中的泛化能力。","openalex","2026-09-11T23:30:29.840009Z"]