Adaptive weight optimization with large pretraining for pest detection

Frequent infestations by agricultural pests reduce crop production and significantly affect economic efficiency. Therefore, timely and effective pest control is critical to improving productivity and facilitate environmental protection. Herein, we propose an adaptive weight optimization method based...

Full description

Saved in:
Bibliographic Details
Main Authors: Kejian Yu, Wenwen Xu, Fuqin Geng, Yunzhi Wu
Format: Article
Language:English
Published: Elsevier 2025-12-01
Series:Ecological Informatics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S157495412500233X
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849233589442445312
author Kejian Yu
Wenwen Xu
Fuqin Geng
Yunzhi Wu
author_facet Kejian Yu
Wenwen Xu
Fuqin Geng
Yunzhi Wu
author_sort Kejian Yu
collection DOAJ
description Frequent infestations by agricultural pests reduce crop production and significantly affect economic efficiency. Therefore, timely and effective pest control is critical to improving productivity and facilitate environmental protection. Herein, we propose an adaptive weight optimization method based on transfer learning for multimodal pest detection. This approach utilizes pretrained model parameters from public datasets to extract features and enhance cross-modal feature from text and images. Accurate pest recognition and localization are achieved through an adaptive loss function, which optimizes the model’s performance across multiple tasks. In tests conducted on IP102 (36 species) and Pest24 (24 species), which are major agricultural pest datasets, the proposed model achieves average precisions of 65.8% and 76.3% at 50% Intersection over Union (IoU) threshold, respectively. By doing so, our model outperforms existing state-of-the-art methods despite using only 30 training cycles. These results highlight the significant practical value of multimodal pest detection methods in enhancing the efficiency and accuracy of agricultural pest identification. The code and dataset are available at https://github.com/Healer-ML/Pest-Detection.
format Article
id doaj-art-098b75964fcd4ff3b3bfd8f70a09b26c
institution Kabale University
issn 1574-9541
language English
publishDate 2025-12-01
publisher Elsevier
record_format Article
series Ecological Informatics
spelling doaj-art-098b75964fcd4ff3b3bfd8f70a09b26c2025-08-20T05:05:10ZengElsevierEcological Informatics1574-95412025-12-019010322410.1016/j.ecoinf.2025.103224Adaptive weight optimization with large pretraining for pest detectionKejian Yu0Wenwen Xu1Fuqin Geng2Yunzhi Wu3School of Information and Artificial Intelligence, Anhui Agricultural University, Hefei 230036, China; Anhui Beidou Precision Agriculture Information Engineering Research Center, Anhui Agricultural University, Hefei 230036, ChinaSchool of Information and Artificial Intelligence, Anhui Agricultural University, Hefei 230036, China; Anhui Beidou Precision Agriculture Information Engineering Research Center, Anhui Agricultural University, Hefei 230036, ChinaSchool of Foreign Languages, Hefei University of Technology, Hefei 230009, ChinaSchool of Information and Artificial Intelligence, Anhui Agricultural University, Hefei 230036, China; Anhui Beidou Precision Agriculture Information Engineering Research Center, Anhui Agricultural University, Hefei 230036, China; Corresponding author at: Anhui Beidou Precision Agriculture Information Engineering Research Center, Anhui Agricultural University, Hefei 230036, China.Frequent infestations by agricultural pests reduce crop production and significantly affect economic efficiency. Therefore, timely and effective pest control is critical to improving productivity and facilitate environmental protection. Herein, we propose an adaptive weight optimization method based on transfer learning for multimodal pest detection. This approach utilizes pretrained model parameters from public datasets to extract features and enhance cross-modal feature from text and images. Accurate pest recognition and localization are achieved through an adaptive loss function, which optimizes the model’s performance across multiple tasks. In tests conducted on IP102 (36 species) and Pest24 (24 species), which are major agricultural pest datasets, the proposed model achieves average precisions of 65.8% and 76.3% at 50% Intersection over Union (IoU) threshold, respectively. By doing so, our model outperforms existing state-of-the-art methods despite using only 30 training cycles. These results highlight the significant practical value of multimodal pest detection methods in enhancing the efficiency and accuracy of agricultural pest identification. The code and dataset are available at https://github.com/Healer-ML/Pest-Detection.http://www.sciencedirect.com/science/article/pii/S157495412500233XPest detectionMulti-modalTransfer learning
spellingShingle Kejian Yu
Wenwen Xu
Fuqin Geng
Yunzhi Wu
Adaptive weight optimization with large pretraining for pest detection
Ecological Informatics
Pest detection
Multi-modal
Transfer learning
title Adaptive weight optimization with large pretraining for pest detection
title_full Adaptive weight optimization with large pretraining for pest detection
title_fullStr Adaptive weight optimization with large pretraining for pest detection
title_full_unstemmed Adaptive weight optimization with large pretraining for pest detection
title_short Adaptive weight optimization with large pretraining for pest detection
title_sort adaptive weight optimization with large pretraining for pest detection
topic Pest detection
Multi-modal
Transfer learning
url http://www.sciencedirect.com/science/article/pii/S157495412500233X
work_keys_str_mv AT kejianyu adaptiveweightoptimizationwithlargepretrainingforpestdetection
AT wenwenxu adaptiveweightoptimizationwithlargepretrainingforpestdetection
AT fuqingeng adaptiveweightoptimizationwithlargepretrainingforpestdetection
AT yunzhiwu adaptiveweightoptimizationwithlargepretrainingforpestdetection