Scholarlink

The enhancement of deep learning techniques within the agricultural sector is a very functional tool in exposing plants to detrimental diseases. The classification and detection of crops based on diseases play a critical function in establishing the rate and criteria of production. In summary, the current models focus on a well-advocated singular method for detecting leaf diseases in Solanum lycopersicum, utilizing sophisticated neural networks to bolster agro-industrial sectors. The innovative architecture currently in use combines the traditional machine learning technique of PCA with a tailored deep neural network referred to as PCA DeepNet, in conjunction with the VGG16 model. Additionally, this integrated framework features a Generative Adversarial Network (GAN) to optimize the data pool. The identification mechanism is reinforced by the F-RCNN. As a result, this methodology achieves a categorization precision of 99.60% and a mean accuracy of 98.04%, yielding a commendable Intersection over Union (IOU) value of 0.94 in recognition and classification. Thus, the proposed approach exceeds all other previously reported futuristic solutions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Tomato Disease Detection Using VGG16 Over PCA DeepNet to Improve Accuracy

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • J. Devadarshini,
  • Carmel Mary Belinda

摘要

The enhancement of deep learning techniques within the agricultural sector is a very functional tool in exposing plants to detrimental diseases. The classification and detection of crops based on diseases play a critical function in establishing the rate and criteria of production. In summary, the current models focus on a well-advocated singular method for detecting leaf diseases in Solanum lycopersicum, utilizing sophisticated neural networks to bolster agro-industrial sectors. The innovative architecture currently in use combines the traditional machine learning technique of PCA with a tailored deep neural network referred to as PCA DeepNet, in conjunction with the VGG16 model. Additionally, this integrated framework features a Generative Adversarial Network (GAN) to optimize the data pool. The identification mechanism is reinforced by the F-RCNN. As a result, this methodology achieves a categorization precision of 99.60% and a mean accuracy of 98.04%, yielding a commendable Intersection over Union (IOU) value of 0.94 in recognition and classification. Thus, the proposed approach exceeds all other previously reported futuristic solutions.