YOLOv5n-MobileNetv4: A Lightweight Crop Pest Detection Algorithm

Qin Yang
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:197-204, 2025.

Abstract

Pests significantly impact crop quality and yield, making their management crucial for agriculture. The real-time and accurate identification of agricultural pests is essential for implementing effective pest control measures, ensuring timely intervention, and minimizing crop losses. Existing pest detection systems face challenges such as low accuracy and excessive parameters, which hinder their efficiency and practicality in real-world applications. Therefore, this paper proposed a real-time pest detector for embedded devices by combining the state-of-the-art mobile device-based MobileNetv4 and lightweight You Only Look Once (YOLO) v5n object detection algorithms, achieving high efficiency and performance. The proposed YOLOv5n-MobileNetv4 model replaced the YOLOv5n backbone with the MobileNetv4 backbone, effectively reducing parameters while maintaining high detection accuracy. Experimental results showed that the improved model achieved 82.1% mAP50 at 87.7 frames per second (FPS). It achieved a 36.2% reduction in parameters and a 31.1% increase in speed, with a slight accuracy drop.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-yang25b, title = {YOLOv5n-MobileNetv4: A Lightweight Crop Pest Detection Algorithm}, author = {Yang, Qin}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {197--204}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/yang25b/yang25b.pdf}, url = {https://proceedings.mlr.press/v278/yang25b.html}, abstract = {Pests significantly impact crop quality and yield, making their management crucial for agriculture. The real-time and accurate identification of agricultural pests is essential for implementing effective pest control measures, ensuring timely intervention, and minimizing crop losses. Existing pest detection systems face challenges such as low accuracy and excessive parameters, which hinder their efficiency and practicality in real-world applications. Therefore, this paper proposed a real-time pest detector for embedded devices by combining the state-of-the-art mobile device-based MobileNetv4 and lightweight You Only Look Once (YOLO) v5n object detection algorithms, achieving high efficiency and performance. The proposed YOLOv5n-MobileNetv4 model replaced the YOLOv5n backbone with the MobileNetv4 backbone, effectively reducing parameters while maintaining high detection accuracy. Experimental results showed that the improved model achieved 82.1% mAP50 at 87.7 frames per second (FPS). It achieved a 36.2% reduction in parameters and a 31.1% increase in speed, with a slight accuracy drop.} }
Endnote
%0 Conference Paper %T YOLOv5n-MobileNetv4: A Lightweight Crop Pest Detection Algorithm %A Qin Yang %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-yang25b %I PMLR %P 197--204 %U https://proceedings.mlr.press/v278/yang25b.html %V 278 %X Pests significantly impact crop quality and yield, making their management crucial for agriculture. The real-time and accurate identification of agricultural pests is essential for implementing effective pest control measures, ensuring timely intervention, and minimizing crop losses. Existing pest detection systems face challenges such as low accuracy and excessive parameters, which hinder their efficiency and practicality in real-world applications. Therefore, this paper proposed a real-time pest detector for embedded devices by combining the state-of-the-art mobile device-based MobileNetv4 and lightweight You Only Look Once (YOLO) v5n object detection algorithms, achieving high efficiency and performance. The proposed YOLOv5n-MobileNetv4 model replaced the YOLOv5n backbone with the MobileNetv4 backbone, effectively reducing parameters while maintaining high detection accuracy. Experimental results showed that the improved model achieved 82.1% mAP50 at 87.7 frames per second (FPS). It achieved a 36.2% reduction in parameters and a 31.1% increase in speed, with a slight accuracy drop.
APA
Yang, Q.. (2025). YOLOv5n-MobileNetv4: A Lightweight Crop Pest Detection Algorithm. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:197-204 Available from https://proceedings.mlr.press/v278/yang25b.html.

Related Material