作为重要的粮食作物,大豆在全球市场中需求巨大。大豆灰斑病严重制约了大豆的优质和高产,因此,快速而准确地识别灰斑病对于大豆种植具有重要意义。目前基于深度学习的大豆灰斑病识别方法存在限制,包括对输入图像尺寸的特定要求、准确率...作为重要的粮食作物,大豆在全球市场中需求巨大。大豆灰斑病严重制约了大豆的优质和高产,因此,快速而准确地识别灰斑病对于大豆种植具有重要意义。目前基于深度学习的大豆灰斑病识别方法存在限制,包括对输入图像尺寸的特定要求、准确率不足,以及模型参数量大导致计算缓慢。针对这些问题,本研究提出了一种改进的VGG16图像识别方法。该方法基于VGG16模型进行了优化,保留了三层卷积层结构,并集成了Inception模块。每个Inception模块由三个并行分支构成:1 × 1卷积层用于降维,1 × 1后接3 × 3卷积层用于特征提取,以及3 × 3最大池化层后接1 × 1卷积层以聚合空间信息。此外,本研究采用全局平均池化层替代了传统的全连接层,减少了模型复杂度并提升计算效率。实验结果表明,该结构不仅放宽了对输入图像尺寸的限制,还显著减少了模型参数,使得参数个数仅为传统VGG16的1.5%,而模型准确率达到99.1%。As an important food crop, soybean is in great demand in the global market. Gray spot of soybean seriously restricts the quality and high yield of soybean, so it is of great significance to identify gray spot quickly and accurately for soybean planting. Current deep learn-based methods for soybean gray spot recognition have limitations, including specific requirements for input image size, inadequate accuracy, and slow calculation due to the large number of model parameters. To solve these problems, an improved VGG16 image recognition method is proposed in this paper. The method is optimized based on the VGG16 model, retains the three-layer convolutional layer structure, and integrates the Inception module. Each Inception module consists of three parallel branches: a 1 × 1 convolutional layer for dimensionality reduction, a 1 × 1 convolutional layer followed by a 3 × 3 convolutional layer for feature extraction, and a 3 × 3 maximum pooling layer followed by a 1 × 1 convolutional layer for aggregation of spatial information. In addition, the global average pooling layer is used to replace the traditional fully connected layer, which reduces the model complexity and improves the computational efficiency. The experimental results show that the structure not only eases the limit on the input image size, but also significantly reduces the model parameters, making the number of parameters only 1.5% of the traditional VGG16, and the model accuracy reaches 99.1%.展开更多
Early and accurate cancer diagnosis through medical imaging is crucial for guiding treatment and enhancing patient survival.However,many state-of-the-art deep learning(DL)methods remain opaque and lack clinical interp...Early and accurate cancer diagnosis through medical imaging is crucial for guiding treatment and enhancing patient survival.However,many state-of-the-art deep learning(DL)methods remain opaque and lack clinical interpretability.This paper presents an explainable artificial intelligence(XAI)framework that combines a fine-tuned Visual Geometry Group 16-layer network(VGG16)convolutional neural network with layer-wise relevance propagation(LRP)to deliver high-performance classification and transparent decision support.This approach is evaluated on the publicly available Kaggle kidney cancer imaging dataset,which comprises labeled cancerous and noncancerous kidney scans.The proposed model achieved 98.75%overall accuracy,with precision,recall,and F1-score each exceeding 98%on an independent test set.Crucially,LRP-derived heatmaps consistently localize anatomically and pathologically significant regions such as tumor margins in agreement with established clinical criteria.The proposed framework enhances clinician trust by delivering pixel-level justifications alongside state-of-the-art predictive performance.It facilitates informed decision-making,thereby addressing a key barrier to the clinical adoption of DL in oncology.展开更多
文摘作为重要的粮食作物,大豆在全球市场中需求巨大。大豆灰斑病严重制约了大豆的优质和高产,因此,快速而准确地识别灰斑病对于大豆种植具有重要意义。目前基于深度学习的大豆灰斑病识别方法存在限制,包括对输入图像尺寸的特定要求、准确率不足,以及模型参数量大导致计算缓慢。针对这些问题,本研究提出了一种改进的VGG16图像识别方法。该方法基于VGG16模型进行了优化,保留了三层卷积层结构,并集成了Inception模块。每个Inception模块由三个并行分支构成:1 × 1卷积层用于降维,1 × 1后接3 × 3卷积层用于特征提取,以及3 × 3最大池化层后接1 × 1卷积层以聚合空间信息。此外,本研究采用全局平均池化层替代了传统的全连接层,减少了模型复杂度并提升计算效率。实验结果表明,该结构不仅放宽了对输入图像尺寸的限制,还显著减少了模型参数,使得参数个数仅为传统VGG16的1.5%,而模型准确率达到99.1%。As an important food crop, soybean is in great demand in the global market. Gray spot of soybean seriously restricts the quality and high yield of soybean, so it is of great significance to identify gray spot quickly and accurately for soybean planting. Current deep learn-based methods for soybean gray spot recognition have limitations, including specific requirements for input image size, inadequate accuracy, and slow calculation due to the large number of model parameters. To solve these problems, an improved VGG16 image recognition method is proposed in this paper. The method is optimized based on the VGG16 model, retains the three-layer convolutional layer structure, and integrates the Inception module. Each Inception module consists of three parallel branches: a 1 × 1 convolutional layer for dimensionality reduction, a 1 × 1 convolutional layer followed by a 3 × 3 convolutional layer for feature extraction, and a 3 × 3 maximum pooling layer followed by a 1 × 1 convolutional layer for aggregation of spatial information. In addition, the global average pooling layer is used to replace the traditional fully connected layer, which reduces the model complexity and improves the computational efficiency. The experimental results show that the structure not only eases the limit on the input image size, but also significantly reduces the model parameters, making the number of parameters only 1.5% of the traditional VGG16, and the model accuracy reaches 99.1%.
基金supported through the Ongoing Research Funding Program(ORF-2025-498),King Saud University,Riyadh,Saudi Arabia.
文摘Early and accurate cancer diagnosis through medical imaging is crucial for guiding treatment and enhancing patient survival.However,many state-of-the-art deep learning(DL)methods remain opaque and lack clinical interpretability.This paper presents an explainable artificial intelligence(XAI)framework that combines a fine-tuned Visual Geometry Group 16-layer network(VGG16)convolutional neural network with layer-wise relevance propagation(LRP)to deliver high-performance classification and transparent decision support.This approach is evaluated on the publicly available Kaggle kidney cancer imaging dataset,which comprises labeled cancerous and noncancerous kidney scans.The proposed model achieved 98.75%overall accuracy,with precision,recall,and F1-score each exceeding 98%on an independent test set.Crucially,LRP-derived heatmaps consistently localize anatomically and pathologically significant regions such as tumor margins in agreement with established clinical criteria.The proposed framework enhances clinician trust by delivering pixel-level justifications alongside state-of-the-art predictive performance.It facilitates informed decision-making,thereby addressing a key barrier to the clinical adoption of DL in oncology.
