亚洲一区欧美在线,日韩欧美视频免费观看,色戒的三场床戏分别是在几段,欧美日韩国产在线人成

基于Mask R-CNN的單株柑橘樹冠識別與分割
作者:
作者單位:

作者簡介:

通訊作者:

中圖分類號:

基金項(xiàng)目:

廣東省重點(diǎn)領(lǐng)域研發(fā)計劃項(xiàng)目(2019B090922001)和江蘇省現(xiàn)代農(nóng)業(yè)裝備與技術(shù)協(xié)同創(chuàng)新中心開放基金項(xiàng)目(4091600016)


Recognition and Segmentation of Individual Citrus Tree Crown Based on Mask R-CNN
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 圖/表
  • |
  • 訪問統(tǒng)計
  • |
  • 參考文獻(xiàn)
  • |
  • 相似文獻(xiàn)
  • |
  • 引證文獻(xiàn)
  • |
  • 資源附件
  • |
  • 文章評論
    摘要:

    針對在復(fù)雜果園背景中難以識別分割單株果樹樹冠的問題,研究了基于Mask R-CNN 神經(jīng)網(wǎng)絡(luò)模型實(shí)現(xiàn)單株柑橘樹冠識別與分割的方法。通過相機(jī)獲取柑橘園圖像數(shù)據(jù),利用Mask R-CNN神經(jīng)網(wǎng)絡(luò)實(shí)現(xiàn)單株柑橘樹冠的識別與分割,根據(jù)測試集的預(yù)測結(jié)果評估模型的性能和可適應(yīng)性,并分析模型的影響因素。結(jié)果表明:參與建模的果園單株樹冠識別分割準(zhǔn)確率為97%,識別時間為0.26s,基本上可滿足果園精準(zhǔn)作業(yè)過程中的樹冠識別要求;未參與建模果園的單株樹冠識別分割準(zhǔn)確率為89%,說明模型對不同品種、不同環(huán)境的果園具有一定的適應(yīng)性;與SegNet模型相比,本文模型準(zhǔn)確率、精確率和召回率均約高5個百分點(diǎn),說明在非目標(biāo)樹冠較多的復(fù)雜果園圖像中具有較好的識別分割效果。本研究可為對靶噴藥、病蟲害防護(hù)、長勢識別與預(yù)估等果園精準(zhǔn)作業(yè)提供重要依據(jù)。

    Abstract:

    The topography of the orchard is variable. The planting density of the fruit trees is large, and the shape of the crown is different. Therefore, it is difficult to recognize the crown of an individual fruit tree in a complex orchard background. A novel method of crown recognition and segmentation based on Mask R-CNN neural network model was studied. The image data of the citrus orchard was obtained through the camera, and the Mask R-CNN neural network was used to realize the recognition and segmentation of the crown of an individual citrus plant. The research results showed that the recognition accuracy of the individual tree crown of the orchard participating in the modeling was 97%, and the recognition time was 0.26s, which can basically meet the requirements of tree crown recognition in the process of precise orchard operation. The recognition accuracy of the single tree crown of the orchard not participating in the modeling was 89%, which showed that the model was suitable for different kinds and environments of orchards. Compared with the SegNet model, the accuracy of the used model was about 5 percentage points higher, indicating that it had a better recognition and segmentation effect in complex orchard images with more non-target tree crowns. Therefore, the recognition and segmentation method can achieve rapid and accurate recognition and segmentation of single tree crown, which provided an important basis for accurate orchard operations such as target spraying, pest protection, growth recognition and prediction.

    參考文獻(xiàn)
    相似文獻(xiàn)
    引證文獻(xiàn)
引用本文

王輝,韓娜娜,呂程序,毛文華,李沐桐,李林.基于Mask R-CNN的單株柑橘樹冠識別與分割[J].農(nóng)業(yè)機(jī)械學(xué)報,2021,52(5):169-174. WANG Hui, HAN Na’na, Lü Chengxu, MAO Wenhua, LI Mutong, LI Lin. Recognition and Segmentation of Individual Citrus Tree Crown Based on Mask R-CNN[J]. Transactions of the Chinese Society for Agricultural Machinery,2021,52(5):169-174.

復(fù)制
分享
文章指標(biāo)
  • 點(diǎn)擊次數(shù):
  • 下載次數(shù):
  • HTML閱讀次數(shù):
  • 引用次數(shù):
歷史
  • 收稿日期:2020-08-04
  • 最后修改日期:
  • 錄用日期:
  • 在線發(fā)布日期: 2021-05-10
  • 出版日期:
文章二維碼