EconPapers    
Economics at your fingertips  
 

Lightweight SCD-YOLOv5s: The Detection of Small Defects on Passion Fruit with Improved YOLOv5s

Yu Zhou, Zhenye Li, Sheng Xue, Min Wu, Tingting Zhu and Chao Ni ()
Additional contact information
Yu Zhou: College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing 210037, China
Zhenye Li: College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing 210037, China
Sheng Xue: College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing 210037, China
Min Wu: College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing 210037, China
Tingting Zhu: College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing 210037, China
Chao Ni: College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing 210037, China

Agriculture, 2025, vol. 15, issue 10, 1-26

Abstract: Accurate detection of surface defects on passion fruits is crucial for maintaining market competitiveness. Numerous small defects present significant challenges for manual inspection. Recently, deep learning (DL) has been widely applied to object detection. In this study, a lightweight neural network, StarC3SE-CBAM-DIoU-YOLOv5s (SCD-YOLOv5s), is proposed based on YOLOv5s for real-time detection of tiny surface defects on passion fruits. Key improvements are introduced as follows: the original C3 module in the backbone is replaced by the enhanced StarC3SE module to achieve a more efficient network structure; the CBAM module is integrated into the neck to improve the extraction of small defect features; and the CIoU loss function is substituted with DIoU-NMS to accelerate convergence and enhance detection accuracy. Experimental results show that SCD-YOLOv5s performs better than YOLOv5s, with precision increased by 13.2%, recall by 1.6%, and F 1 - score by 17.0%. Additionally, improvements of 6.7% in mAP@0.5 and 5.5% in mAP@0.95 are observed. Compared with manual detection, the proposed model enhances detection efficiency by reducing errors caused by subjective judgment. It also achieves faster inference speed (26.66 FPS), and reductions of 9.6% in parameters and 8.6% in weight size, while maintaining high detection performance. These results indicate that SCD-YOLOv5s is effective for defect detection in agricultural applications.

Keywords: passion fruit; defect detection; YOLOv5s; attention mechanism (search for similar items in EconPapers)
JEL-codes: Q1 Q10 Q11 Q12 Q13 Q14 Q15 Q16 Q17 Q18 (search for similar items in EconPapers)
Date: 2025
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2077-0472/15/10/1111/pdf (application/pdf)
https://www.mdpi.com/2077-0472/15/10/1111/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jagris:v:15:y:2025:i:10:p:1111-:d:1661216

Access Statistics for this article

Agriculture is currently edited by Ms. Leda Xuan

More articles in Agriculture from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().

 
Page updated 2025-06-07
Handle: RePEc:gam:jagris:v:15:y:2025:i:10:p:1111-:d:1661216