生態農業：農業地景生態監測及復育【土木工程學系/楊明德特聘教授】
| 論文篇名 | 英文：Rice Seedling Detection in UAV Images Using Transfer Learning and Machine Learning 中文：使用遷移學習和機器學習在無人機圖像中檢測水稻幼苗 |
| 期刊名稱 | REMOTE SENSING |
| 發表年份,卷數,起迄頁數 | 2022, 14(12), 2837 |
| 作者 | Tseng, Hsin-Hung(曾信鴻); Yang, Ming-Der(楊明德)*; Saminathan, R.; Hsu, Yu-Chun(許鈺群); Yang, Chin-Ying; Wu, Dong-Hong |
| DOI | 10.3390/rs14122837 |
| 中文摘要 | 為了滿足對農產品的需求，近來研究多專注於精準農業，以降低投入並增加作物產量。基於計算機視覺和無人機 (UAV) 獲取影像的作物檢測在精準農業中發揮著至關重要的作用。近年來，機器學習已成功應用於影像處理，如影像分類、檢測和分割。因此，本研究的目的是使用兩種機器學習模型 EfficientDet-D0 和 Faster RCNN 的遷移學習檢測稻田中的水稻幼苗，並將結果與傳統方法——基於定向梯度直方圖 (HOG) 的方法進行比較。這項研究依靠無人機影像數據集建立一個模型來檢測微小的水稻幼苗。 HOG-SVM 分類器經過訓練，在訓練和測試中都達到了 99% 的 F1 分數。 HOG-SVM、EfficientDet 和 Faster R-CNN 模型的性能分別以平均精度 (mAP) 衡量，在訓練中分別為 70.0%、95.5% 和幾乎 100%，在測試中分別為 70.2%、83.2% 和 88.8% ，以及平均交並比 (mIoU)，訓練中分別為 46.5%、67.6% 和 99.6%，測試中分別為 46.6%、57.5% 和 63.7%。這三個模型還使用不同日期獲取的三個附加數據集進行測試，以評估模型在各種成像條件下的適用性。結果表明，兩種基於 CNN 的模型都優於 HOG-SVM，mAP 和 mIoU 高出 10%。此外，計算速度至少比具有滑動視窗的 HOG-SVM 快 1000 倍。總體而言，採用遷移學習可以快速建立具有良好性能的物件檢測程序。 |
| 英文摘要 | To meet demand for agriculture products, researchers have recently focused on precision agriculture to increase crop production with less input. Crop detection based on computer vision with unmanned aerial vehicle (UAV)-acquired images plays a vital role in precision agriculture. In recent years, machine learning has been successfully applied in image processing for classification, detection and segmentation. Accordingly, the aim of this study is to detect rice seedlings in paddy fields using transfer learning from two machine learning models, EfficientDet-D0 and Faster RCNN, and to compare the results to the legacy approach—histograms of oriented gradients (HOG)-based support vector machine (SVM) classification. This study relies on a significant UAV image dataset to build a model to detect tiny rice seedlings. The HOG-SVM classifier was trained and achieved an F1-score of 99% in both training and testing. The performance of HOG-SVM, EfficientDet and Faster R-CNN models, respectively, were measured in mean average precision (mAP), with 70.0%, 95.5% and almost 100% in training and 70.2%, 83.2% and 88.8% in testing, and mean Intersection-over-Union (mIoU), with 46.5%, 67.6% and 99.6% in training and 46.6%, 57.5% and 63.7% in testing. The three models were also measured with three additional datasets acquired on different dates to evaluate model applicability with various imaging conditions. The results demonstrate that both CNN-based models outperform HOG-SVM, with a 10% higher mAP and mIoU. Further, computation speed is at least 1000 times faster than that of HOG-SVM with sliding window. Overall, the adoption of transfer learning allows for rapid establishment of object detection applications with promising performance. |
| 發表成果與本中心研究主題相關性 | 農地的人工智慧辨識。 |
