An Improved YOLOX Algorithm for Forest Insect Pest Detection
Open Access
- 23 August 2022
- journal article
- research article
- Published by Wiley in Computational Intelligence and Neuroscience
- Vol. 2022, 1-12
Abstract
A large number of insect pests in the forest will seriously affect the construction of forest resources and agriculture in China. In this regard, in order to deeply understand and analyze the existing forest pest detection technology, it is found that it cannot meet practical needs. In order to prevent the harm caused by forest pests, it is necessary to correctly identify the types of pests and take targeted control measures. Therefore, this paper proposes a forest pest detection algorithm based on improved YOLOX. Firstly, aiming at the problem that there are few image data of real deep forest pests in the wild, we use Mosaic, Mixup, and random erasure data enhancement to preprocess the images. Secondly, in order to extract fine-grained features, shallow information is introduced into the existing network architecture, and a two-way cross-scale feature fusion mechanism is adopted. Finally, the improved YOLOX algorithm proposed in this paper has achieved the best results on the public forest pest dataset IP102.Keywords
Funding Information
- National Natural Science Foundation of China (61876168, 2022LSDMIS02)
- Ministry of Education of the People's Republic of China (61876168, 2022LSDMIS02)
This publication has 11 references indexed in Scilit:
- Identification of crop diseases using improved convolutional neural networksIET Computer Vision, 2020
- Cotton pests classification in field-based images using deep residual networksComputers and Electronics in Agriculture, 2020
- Detection of stored-grain insects using deep learningComputers and Electronics in Agriculture, 2018
- Pest identification via deep residual learning in complex backgroundComputers and Electronics in Agriculture, 2017
- Pedestrian detection based on gradient and texture feature integrationNeurocomputing, 2017
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal NetworksIEEE Transactions on Pattern Analysis and Machine Intelligence, 2016
- Detection of aphids in wheat fields using a computer vision techniqueBiosystems Engineering, 2016
- ML-KNN: A lazy learning approach to multi-label learningPattern Recognition, 2007
- Distinctive Image Features from Scale-Invariant KeypointsInternational Journal of Computer Vision, 2004
- Support-vector networksMachine Learning, 1995