Megaweeds: an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset

Sophie Wildeboer; Jurrian Doornbos; Önder Babur; Kwabena E. Bennin

doi:10.1007/978-3-032-15375-3_11

Megaweeds: an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset

Sophie Wildeboer
, Jurrian Doornbos
, Önder Babur (Corresponding author-nrf)
, Kwabena E. Bennin

Software Engineering and Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

Modern agriculture must address growing food demand, labor shortages, and environmental concerns. Precision agriculture (PA), which involves crop monitoring and automation, offers solutions. A key element of PA is crop and weed detection, often achieved with deep learning. This study applies YOLOv7, an object detection model, to this task. While most approaches pre-train YOLOv7 on the general COCO dataset, COCO lacks plant images. Therefore, we built the MegaWeeds (MW) dataset, which contains seven existing datasets for weed and crop detection with a total of 18,287 images (and 19,317 label files). Performance was tested using the Lincolnbeet dataset of sugar beets and weeds, comparing models pre-trained on COCO and MW. Results showed similar mean average precision scores, i.e. 78.7% (COCO pretrained) and 78.6% (MW pretrained). Nonetheless, pretraining with MW proved advantageous by enabling faster convergence and better zero-shot detection than COCO. The study surpassed earlier YOLOv7 results on the same dataset and demonstrated that domain-specific pretraining shortens transfer learning and enhances detection efficiency.

Original language	English
Title of host publication	Innovative Agricultural Technologies
Subtitle of host publication	Proceedings of IAT Congress 2025
Editors	Longsheng Fu, Jitendra Paliwal, Hasan H. Silleli, Barbara Sturm, Fernando Auat Cheein
Place of Publication	Cham
Publisher	Springer
Pages	136-156
Number of pages	21
ISBN (Electronic)	978-3-032-15375-3
ISBN (Print)	978-3-032-15374-6, 978-3-032-15377-7
DOIs	https://doi.org/10.1007/978-3-032-15375-3_11
Publication status	Published - 31 Jan 2026
Event	15th International Congress of the Innovative Agricultural Technologies, IAT 2025 - Antalya, Turkey Duration: 15 Oct 2025 → 19 Oct 2025

Publication series

Name	Lecture Notes in Civil Engineering (LNCE)
Volume	805
ISSN (Print)	2366-2557
ISSN (Electronic)	2366-2565

Conference

Conference	15th International Congress of the Innovative Agricultural Technologies, IAT 2025
Country/Territory	Turkey
City	Antalya
Period	15/10/25 → 19/10/25

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Keywords

Crop Detection
Deep Learning
Precision Agriculture
Weed Detection
YOLOv7

Access to Document

10.1007/978-3-032-15375-3_11

Cite this

Wildeboer, S., Doornbos, J., Babur, Ö., & Bennin, K. E. (2026). Megaweeds: an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset. In L. Fu, J. Paliwal, H. H. Silleli, B. Sturm, & F. Auat Cheein (Eds.), Innovative Agricultural Technologies: Proceedings of IAT Congress 2025 (pp. 136-156). (Lecture Notes in Civil Engineering (LNCE); Vol. 805). Springer. https://doi.org/10.1007/978-3-032-15375-3_11

Wildeboer, Sophie ; Doornbos, Jurrian ; Babur, Önder et al. / Megaweeds : an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset. Innovative Agricultural Technologies: Proceedings of IAT Congress 2025. editor / Longsheng Fu ; Jitendra Paliwal ; Hasan H. Silleli ; Barbara Sturm ; Fernando Auat Cheein. Cham : Springer, 2026. pp. 136-156 (Lecture Notes in Civil Engineering (LNCE)).

@inproceedings{1689ba57b0f742c496ef38f7b46259ca,

title = "Megaweeds: an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset",

abstract = "Modern agriculture must address growing food demand, labor shortages, and environmental concerns. Precision agriculture (PA), which involves crop monitoring and automation, offers solutions. A key element of PA is crop and weed detection, often achieved with deep learning. This study applies YOLOv7, an object detection model, to this task. While most approaches pre-train YOLOv7 on the general COCO dataset, COCO lacks plant images. Therefore, we built the MegaWeeds (MW) dataset, which contains seven existing datasets for weed and crop detection with a total of 18,287 images (and 19,317 label files). Performance was tested using the Lincolnbeet dataset of sugar beets and weeds, comparing models pre-trained on COCO and MW. Results showed similar mean average precision scores, i.e. 78.7\% (COCO pretrained) and 78.6\% (MW pretrained). Nonetheless, pretraining with MW proved advantageous by enabling faster convergence and better zero-shot detection than COCO. The study surpassed earlier YOLOv7 results on the same dataset and demonstrated that domain-specific pretraining shortens transfer learning and enhances detection efficiency.",

keywords = "Crop Detection, Deep Learning, Precision Agriculture, Weed Detection, YOLOv7",

author = "Sophie Wildeboer and Jurrian Doornbos and {\"O}nder Babur and Bennin, \{Kwabena E.\}",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.; 15th International Congress of the Innovative Agricultural Technologies, IAT 2025 ; Conference date: 15-10-2025 Through 19-10-2025",

year = "2026",

month = jan,

day = "31",

doi = "10.1007/978-3-032-15375-3\_11",

language = "English",

isbn = "978-3-032-15374-6",

series = "Lecture Notes in Civil Engineering (LNCE)",

publisher = "Springer",

pages = "136--156",

editor = "Longsheng Fu and Jitendra Paliwal and Silleli, \{Hasan H.\} and Barbara Sturm and \{Auat Cheein\}, Fernando",

booktitle = "Innovative Agricultural Technologies",

address = "Germany",

}

Wildeboer, S, Doornbos, J, Babur, Ö & Bennin, KE 2026, Megaweeds: an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset. in L Fu, J Paliwal, HH Silleli, B Sturm & F Auat Cheein (eds), Innovative Agricultural Technologies: Proceedings of IAT Congress 2025. Lecture Notes in Civil Engineering (LNCE), vol. 805, Springer, Cham, pp. 136-156, 15th International Congress of the Innovative Agricultural Technologies, IAT 2025, Antalya, Turkey, 15/10/25. https://doi.org/10.1007/978-3-032-15375-3_11

Megaweeds: an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset. / Wildeboer, Sophie; Doornbos, Jurrian; Babur, Önder (Corresponding author-nrf) et al.
Innovative Agricultural Technologies: Proceedings of IAT Congress 2025. ed. / Longsheng Fu; Jitendra Paliwal; Hasan H. Silleli; Barbara Sturm; Fernando Auat Cheein. Cham: Springer, 2026. p. 136-156 (Lecture Notes in Civil Engineering (LNCE); Vol. 805).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Megaweeds

T2 - 15th International Congress of the Innovative Agricultural Technologies, IAT 2025

AU - Wildeboer, Sophie

AU - Doornbos, Jurrian

AU - Bennin, Kwabena E.

A2 - Babur, Önder

A2 - Fu, Longsheng

A2 - Paliwal, Jitendra

A2 - Silleli, Hasan H.

A2 - Sturm, Barbara

A2 - Auat Cheein, Fernando

PY - 2026/1/31

Y1 - 2026/1/31

N2 - Modern agriculture must address growing food demand, labor shortages, and environmental concerns. Precision agriculture (PA), which involves crop monitoring and automation, offers solutions. A key element of PA is crop and weed detection, often achieved with deep learning. This study applies YOLOv7, an object detection model, to this task. While most approaches pre-train YOLOv7 on the general COCO dataset, COCO lacks plant images. Therefore, we built the MegaWeeds (MW) dataset, which contains seven existing datasets for weed and crop detection with a total of 18,287 images (and 19,317 label files). Performance was tested using the Lincolnbeet dataset of sugar beets and weeds, comparing models pre-trained on COCO and MW. Results showed similar mean average precision scores, i.e. 78.7% (COCO pretrained) and 78.6% (MW pretrained). Nonetheless, pretraining with MW proved advantageous by enabling faster convergence and better zero-shot detection than COCO. The study surpassed earlier YOLOv7 results on the same dataset and demonstrated that domain-specific pretraining shortens transfer learning and enhances detection efficiency.

AB - Modern agriculture must address growing food demand, labor shortages, and environmental concerns. Precision agriculture (PA), which involves crop monitoring and automation, offers solutions. A key element of PA is crop and weed detection, often achieved with deep learning. This study applies YOLOv7, an object detection model, to this task. While most approaches pre-train YOLOv7 on the general COCO dataset, COCO lacks plant images. Therefore, we built the MegaWeeds (MW) dataset, which contains seven existing datasets for weed and crop detection with a total of 18,287 images (and 19,317 label files). Performance was tested using the Lincolnbeet dataset of sugar beets and weeds, comparing models pre-trained on COCO and MW. Results showed similar mean average precision scores, i.e. 78.7% (COCO pretrained) and 78.6% (MW pretrained). Nonetheless, pretraining with MW proved advantageous by enabling faster convergence and better zero-shot detection than COCO. The study surpassed earlier YOLOv7 results on the same dataset and demonstrated that domain-specific pretraining shortens transfer learning and enhances detection efficiency.

KW - Crop Detection

KW - Deep Learning

KW - Precision Agriculture

KW - Weed Detection

KW - YOLOv7

UR - https://www.scopus.com/pages/publications/105029236621

U2 - 10.1007/978-3-032-15375-3_11

DO - 10.1007/978-3-032-15375-3_11

M3 - Conference contribution

AN - SCOPUS:105029236621

SN - 978-3-032-15374-6

SN - 978-3-032-15377-7

T3 - Lecture Notes in Civil Engineering (LNCE)

SP - 136

EP - 156

BT - Innovative Agricultural Technologies

PB - Springer

CY - Cham

Y2 - 15 October 2025 through 19 October 2025

ER -

Wildeboer S, Doornbos J, Babur Ö, Bennin KE. Megaweeds: an Experimental Study on Weed Detection with YOLOv7 Using a Novel Dataset. In Fu L, Paliwal J, Silleli HH, Sturm B, Auat Cheein F, editors, Innovative Agricultural Technologies: Proceedings of IAT Congress 2025. Cham: Springer. 2026. p. 136-156. (Lecture Notes in Civil Engineering (LNCE)). doi: 10.1007/978-3-032-15375-3_11