Enhanced Detection of Soybean Leaf Diseases Using an Improved Yolov5 Model
Journal Title: International Journal of Knowledge and Innovation Studies - Year 2024, Vol 2, Issue 1
Abstract
To facilitate early intervention and control efforts, this study proposes a soybean leaf disease detection method based on an improved Yolov5 model. Initially, image preprocessing is applied to two datasets of diseased soybean leaf images. Subsequently, the original Yolov5s network model is modified by replacing the Spatial Pyramid Pooling (SPP) module with a simplified SimSPPF for more efficient and precise feature extraction. The backbone Convolutional Neural Network (CNN) is enhanced with the Bottleneck transformer (BotNet) self-attention mechanism to accelerate detection speed. The Complete Intersection over Union (CIoU) loss function is replaced by EIoU-Loss to increase the model's inference speed, and Enhanced Intersection over Union (EIoU)-Non-Maximum Suppression (NMS) is used instead of traditional NMS to optimize the handling of prediction boxes. Experimental results demonstrate that the modified Yolov5s model increases the mean Average Precision (mAP) value by 4.5% compared to the original Yolov5 network model for the detection and identification of soybean leaf diseases. Therefore, the proposed method effectively detects and identifies soybean leaf diseases and can be validated for practicality in actual production environments.
Authors and Affiliations
Shiqin Peng, Guiqing Xi, Yongshun Wei, Ling Yu
Keywords
Related Articles
- EP ID EP744651
- DOI https://doi.org/10.56578/ijkis020105
- Views 26
- Downloads 0
Application of Complex Polytopic Fuzzy Information Systems in Knowledge Engineering: Decision Support for COVID-19 Vaccine Selection
This paper aims to introduce the concepts of complex Polytopic fuzzy sets (CPoFSs) and complex Polytopic fuzzy numbers (CPoFNs), advancing the field of fuzzy logic. Three innovative aggregation operators based on CPoFNs...
Enhancement of the Defining Interrelationships Between Ranked Criteria II Method Using Interval Grey Numbers for Application in the Grey-Rough MCDM Model
Multi-Criteria Decision-Making (MCDM) represents a critical area of research, particularly in artificial intelligence, through the modeling of real-world decision-making scenarios. Numerous methods have been developed to...
Evaluating the Knowledge Economies within the European Union: A Global Knowledge Index Ranking via Entropy and CRADIS Methodologies
In this study, a novel methodology is proposed for ranking the knowledge economies of European Union (EU) countries, leveraging their positioning within the global knowledge index (GKI). The GKI, encompassing seven pivot...
Generalized and Group-Generalized Parameter Based Fermatean Fuzzy Aggregation Operators with Application to Decision-Making
Fermatean fuzzy set (FRFS) is very helpful in representing vague information that occurs in real world circumstances. Their eminent characteristic of FRFS is that the degree of membership ℑℓ and degree of non-membership...
Enhanced Global Image Segmentation: Addressing Pixel Inhomogeneity and Noise with Average Convolution and Entropy-Based Local Factor
In the field of computer vision and digital image processing, the division of images into meaningful segments is a pivotal task. This paper introduces an innovative global image segmentation model, distinguished for its...