東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Near Infrared Technology Application...

Yamada, William,

FindBook

Google Book

Amazon

博客來

Near Infrared Technology Applications in Agriculture /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Near Infrared Technology Applications in Agriculture // William Yamada.
作者:	Yamada, William,
面頁冊數:	1 electronic resource (158 pages)
附註:	Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
Contained By:	Dissertations Abstracts International86-12B.
標題:	Agriculture. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32116351
ISBN:	9798280771109

Near Infrared Technology Applications in Agriculture /
Yamada, William,

Near Infrared Technology Applications in Agriculture /William Yamada. - 1 electronic resource (158 pages)

Source: Dissertations Abstracts International, Volume: 86-12, Section: B.

Precision agriculture demands rapid, non-invasive tools to monitor crop quality and minimize losses, especially during harvest. Traditional laboratory-based analyses, such as wet chemistry and mechanical sieving, are accurate but slow, labor-intensive, and impractical for in-field deployment. This dissertation presents machine learning frameworks that integrate near-infrared (NIR) spectroscopy, self-supervised computer vision, and multimodal data fusion to estimate forage quality and detect pre-harvest losses in real time.First, handheld NIR spectrometers were evaluated for their ability to estimate nutritional traits in undried forage. Through spectral preprocessing and calibration transfer, the study demonstrated that portable devices can yield accurate predictions of crude protein and digestibility, offering viable alternatives to benchtop systems for on-farm use.Second, the dissertation addresses grain loss monitoring in row crops by deploying deep learning detectors (YOLO, DETR, YOLOv8) to identify uncollected grain from ground-based RGB-D imagery. These models, trained and validated on wheat, oats, and soybean datasets, demonstrated strong object-level performance and spatial loss quantification, enabling automation of a process that traditionally relies on manual collection and counting.Third, a late fusion framework was introduced to combine RGB-derived morphometrics with NIR spectral features selected via Partial Least Squares Regression and Variable Importance in Projection (VIP). This approach yielded a Random Forest model achieving a Pearson correlation of r = 0.968 and RMSECV = 2.16%, outperforming previous single-modality methods and enabling practical, edge-compatible deployment.Finally, RGB images of corn silage were processed using self-supervised Vision Transformers (DINOv2), enabling regression of the Corn Silage Processing Score (CSPS) without manual annotation or kernel segmentation. Performance was further improved using a proxy variable-the proportion of particles above the 4.75 mm sieve (FoG)-which enhanced interpretability and predictive accuracy.Collectively, this research advances the development of interpretable, deployable, and multimodal sensing systems for agriculture. It demonstrates that computer vision and spectroscopy can be fused to assess silage quality and crop losses accurately, bridging the gap between lab-grade analytics and in-field decision-making for harvest optimization.

English

ISBN: 9798280771109Subjects--Topical Terms:

518588
Agriculture.
Subjects--Index Terms:

Agricultural machinery

Near Infrared Technology Applications in Agriculture /
LDR:03965nmm a22004573i 4500 001 2435881
005 20251029060413.5
006 m o d
007 cr|nu||||||||
008 261011s2025 miu||||||m |||||||eng d
020 $a 9798280771109
035 $a (MiAaPQD)AAI32116351
035 $a AAI32116351
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Yamada, William, $e author. $3 3826354
245 1 0 $a Near Infrared Technology Applications in Agriculture / $c William Yamada.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (158 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
500 $a Advisors: Digman, Matthew F. Committee members: Runge, Troy; Zhang, Zhou; Dorea, Joao.
502 $b Ph.D. $c The University of Wisconsin - Madison $d 2025.
520 $a Precision agriculture demands rapid, non-invasive tools to monitor crop quality and minimize losses, especially during harvest. Traditional laboratory-based analyses, such as wet chemistry and mechanical sieving, are accurate but slow, labor-intensive, and impractical for in-field deployment. This dissertation presents machine learning frameworks that integrate near-infrared (NIR) spectroscopy, self-supervised computer vision, and multimodal data fusion to estimate forage quality and detect pre-harvest losses in real time.First, handheld NIR spectrometers were evaluated for their ability to estimate nutritional traits in undried forage. Through spectral preprocessing and calibration transfer, the study demonstrated that portable devices can yield accurate predictions of crude protein and digestibility, offering viable alternatives to benchtop systems for on-farm use.Second, the dissertation addresses grain loss monitoring in row crops by deploying deep learning detectors (YOLO, DETR, YOLOv8) to identify uncollected grain from ground-based RGB-D imagery. These models, trained and validated on wheat, oats, and soybean datasets, demonstrated strong object-level performance and spatial loss quantification, enabling automation of a process that traditionally relies on manual collection and counting.Third, a late fusion framework was introduced to combine RGB-derived morphometrics with NIR spectral features selected via Partial Least Squares Regression and Variable Importance in Projection (VIP). This approach yielded a Random Forest model achieving a Pearson correlation of r = 0.968 and RMSECV = 2.16%, outperforming previous single-modality methods and enabling practical, edge-compatible deployment.Finally, RGB images of corn silage were processed using self-supervised Vision Transformers (DINOv2), enabling regression of the Corn Silage Processing Score (CSPS) without manual annotation or kernel segmentation. Performance was further improved using a proxy variable-the proportion of particles above the 4.75 mm sieve (FoG)-which enhanced interpretability and predictive accuracy.Collectively, this research advances the development of interpretable, deployable, and multimodal sensing systems for agriculture. It demonstrates that computer vision and spectroscopy can be fused to assess silage quality and crop losses accurately, bridging the gap between lab-grade analytics and in-field decision-making for harvest optimization.
546 $a English
590 $a School code: 0262
650 4 $a Agriculture. $3 518588
650 4 $a Computer science. $3 523869
650 4 $a Engineering. $3 586835
650 4 $a Agricultural engineering. $3 3168406
653 $a Agricultural machinery
653 $a Machine learning
653 $a Near-infrared spectroscopy
653 $a Precision agriculture
653 $a Corn Silage Processing Score
690 $a 0473
690 $a 0984
690 $a 0537
690 $a 0800
690 $a 0539
710 2 $a The University of Wisconsin - Madison. $b Biological Systems Engineering. $e degree granting institution. $3 3826355
720 1 $a Digman, Matthew F. $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-12B.
790 $a 0262
791 $a Ph.D.
792 $a 2025
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32116351