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

基于少量標(biāo)注樣本的茶芽目標(biāo)檢測YSVD-Tea算法
作者:
作者單位:

作者簡介:

通訊作者:

中圖分類號:

基金項目:

國家重點研發(fā)計劃項目(2021YFD1601102)和安徽省自然科學(xué)基金項目(2308085MC84)


YSVD-Tea Algorithm for Tea Bud Object Detection Based on Few Annotated Samples
Author:
Affiliation:

Fund Project:

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

    構(gòu)建大規(guī)模茶芽目標(biāo)檢測數(shù)據(jù)集是一項耗時且繁瑣的任務(wù),為了降低數(shù)據(jù)集構(gòu)建成本,探索少量標(biāo)注樣本的算法尤為必要。本文提出了YSVD-Tea (YOLO singular value decomposition for tea bud detection)算法,通過將預(yù)訓(xùn)練模型中的基礎(chǔ)卷積替換為3個連續(xù)的矩陣結(jié)構(gòu),實現(xiàn)了對YOLOX算法結(jié)構(gòu)的重構(gòu)。通過維度變化和奇異值分解操作,將預(yù)訓(xùn)練權(quán)重轉(zhuǎn)換為與重構(gòu)算法結(jié)構(gòu)相對應(yīng)的權(quán)重,從而將需要進行遷移學(xué)習(xí)的權(quán)重和需要保留的權(quán)重分離開,實現(xiàn)保留預(yù)訓(xùn)練模型先驗信息的目的。在3種不同數(shù)量的數(shù)據(jù)集上分別進行了訓(xùn)練和驗證。在最小數(shù)量的1/3數(shù)據(jù)集上,YSVD-Tea算法相較于改進前的YOLOX算法,mAP提高20.3個百分點。對比測試集與訓(xùn)練集的性能指標(biāo),YSVD-Tea算法在測試集與訓(xùn)練集的mAP差距僅為21.9%,明顯小于YOLOX的40.6%和Faster R-CNN的55.4%。在數(shù)量最大的數(shù)據(jù)集上,YOLOX算法精確率、召回率、F1值、mAP分別為86.4%、87.0%、86.7%和88.3%,相較于對比算法均最高。YSVD-Tea在保證良好性能的同時,能夠更好地適應(yīng)少量標(biāo)注樣本的茶芽目標(biāo)檢測任務(wù)。

    Abstract:

    Constructing a large-scale dataset for tea bud object detection is a time-consuming and intricate task. To mitigate the cost of dataset construction, exploring algorithms with a minimal number of annotated samples is particularly necessary. The YOLO singular value decomposition for tea bud detection (YSVD-Tea) algorithm was introduced, which achieved the reconstruction of the YOLOX structure by replacing the basic convolution in the pre-trained model with three consecutive matrix structures. Through dimension transformation and singular value decomposition operations, pre-trained weights were converted into weights corresponding to the reconstructed algorithm structure, thereby separating the weights that require transfer learning from those that needed to be retained. This achieved the goal of preserving the general semantic information of the pre-trained model. Training and validation on three datasets of varying sizes were conducted. On the smallest 1/3 dataset, the YSVD-Tea algorithm showed a 20.3 percentage points improvement in mAP compared with the original YOLOX algorithm. Comparing performance metrics between the test and training sets, the mAP difference for the YSVD-Tea algorithm was only 21.9%, which was significantly lower than YOLOX’s 40.6% and Faster R-CNN’s 55.4%. In training with the largest complete dataset, the YOLOX algorithm achieved precision, recall, F1 score, and mAP of 86.4%, 87.0%, 86.7%, and 88.3%, respectively, surpassing the comparison algorithms. YSVD-Tea algorithm demonstrated superior suitability for the task of tea bud object detection, especially when confronted with a limited number of annotated samples.

    參考文獻
    相似文獻
    引證文獻
引用本文

鄭子秋,宋彥,陳霖,張航,寧井銘.基于少量標(biāo)注樣本的茶芽目標(biāo)檢測YSVD-Tea算法[J].農(nóng)業(yè)機械學(xué)報,2024,55(8):301-311. ZHENG Ziqiu, SONG Yan, CHEN Lin, ZHANG Hang, NING Jingming. YSVD-Tea Algorithm for Tea Bud Object Detection Based on Few Annotated Samples[J]. Transactions of the Chinese Society for Agricultural Machinery,2024,55(8):301-311.

復(fù)制
分享
文章指標(biāo)
  • 點擊次數(shù):
  • 下載次數(shù):
  • HTML閱讀次數(shù):
  • 引用次數(shù):
歷史
  • 收稿日期:2023-11-23
  • 最后修改日期:
  • 錄用日期:
  • 在線發(fā)布日期: 2024-08-10
  • 出版日期: