Scholarlink

Turmeric is a widely used spice due to its medicinal, culinary, and dyeing properties. However, its frequent adulteration poses significant health risks, including heart disease, metabolic disorders, and even death. Therefore, a swift, dependable, and non-destructive method is imperative. This research paper aims to detect adulterants in turmeric using thermal and RGB images. Pure turmeric powder is blended with 20% chalk powder, chickpea powder, and starch, along with 10% tartrazine adulterants. The prepared samples undergo heating at 50 and 70  \(^\circ \) C, while thermal and RGB images are captured. Applying CNN and VGG16 models, we achieve accuracies of 77.11% and 63.46% on CNN, and 93.08% and 95.08% on VGG16 for thermal images at 50 and 70  \(^\circ \) C, respectively. For RGB, we attain accuracies of 68.18% on CNN and 95.45% on VGG16. These results indicate the viability of detecting adulteration using thermal images, with VGG16 demonstrating superior performance.

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

Machine Learning Based Approach to Detect Adulteration in Turmeric Using RGB and Thermal Images

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Rupinder Kaur,
  • Simrandeep Singh,
  • Mukesh Saini

摘要

Turmeric is a widely used spice due to its medicinal, culinary, and dyeing properties. However, its frequent adulteration poses significant health risks, including heart disease, metabolic disorders, and even death. Therefore, a swift, dependable, and non-destructive method is imperative. This research paper aims to detect adulterants in turmeric using thermal and RGB images. Pure turmeric powder is blended with 20% chalk powder, chickpea powder, and starch, along with 10% tartrazine adulterants. The prepared samples undergo heating at 50 and 70  \(^\circ \) C, while thermal and RGB images are captured. Applying CNN and VGG16 models, we achieve accuracies of 77.11% and 63.46% on CNN, and 93.08% and 95.08% on VGG16 for thermal images at 50 and 70  \(^\circ \) C, respectively. For RGB, we attain accuracies of 68.18% on CNN and 95.45% on VGG16. These results indicate the viability of detecting adulteration using thermal images, with VGG16 demonstrating superior performance.