View in EDS PDF Full Text

A Lightweight Improved YOLOv8-Based Method for Mushroom Maturity Stage Detection.

Saved in:
Bibliographic Details
Title: A Lightweight Improved YOLOv8-Based Method for Mushroom Maturity Stage Detection.
Authors: Wu, Hao1 y00452505127@njit.edu.cn, Song, Chengcheng2 schengcheng@126.com, Zhou, Baoyu1 y00452505136@njit.edu.cn, Chen, Bo1 njit_airobot@126.com, Shang, Yanling3 hnnhsyl@126.com, Gao, Fangzheng4 gaofz@126.com
Source: IAENG International Journal of Computer Science. Aug2026, Vol. 53 Issue 8, p3135-3143. 9p.
Subjects: Object recognition (Computer vision), Embedded computer systems, Machine learning
Abstract: This paper proposes DPI-YOLO, a lightweight model based on improved YOLOv8n, for automatic mushroom maturity detection. To address low harvesting efficiency, high labor cost, and edge deployment requirements, the network is optimized in accuracy and complexity. PConv is introduced into the detection head to reduce redundant computation, DySample is used to enhance multi-scale feature fusion, and Inner-CIoU is adopted to improve bounding box localization. Experimental results show that DPI-YOLO achieves 83.0% mAP50, 83.6% Precision, and 75.0% Recall, with 2.43M parameters and 5.6 GFLOPs. Compared with YOLOv8n, mAP50 increases by 1.4% while model complexity decreases. The TensorRT deployment on Jetson Orin Nano Super achieves a total processing time of about 26.9 ms per image, confirming its feasibility for embedded mushroom maturity detection. [ABSTRACT FROM AUTHOR]
Copyright of IAENG International Journal of Computer Science is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:This paper proposes DPI-YOLO, a lightweight model based on improved YOLOv8n, for automatic mushroom maturity detection. To address low harvesting efficiency, high labor cost, and edge deployment requirements, the network is optimized in accuracy and complexity. PConv is introduced into the detection head to reduce redundant computation, DySample is used to enhance multi-scale feature fusion, and Inner-CIoU is adopted to improve bounding box localization. Experimental results show that DPI-YOLO achieves 83.0% mAP50, 83.6% Precision, and 75.0% Recall, with 2.43M parameters and 5.6 GFLOPs. Compared with YOLOv8n, mAP50 increases by 1.4% while model complexity decreases. The TensorRT deployment on Jetson Orin Nano Super achieves a total processing time of about 26.9 ms per image, confirming its feasibility for embedded mushroom maturity detection. [ABSTRACT FROM AUTHOR]
ISSN:1819656X