Paddy seed variety classification using transfer learning based on deep learning
Deachrut Jaithavil, Sirichai Triamlumlerd, Manoch Pracha
Abstract
Classification of paddy seed varieties is a critical task that determines the price and quality of rice in the market. Deep learning methods that have been successfully applied in different areas are not even in image classification. In this paper, we proposed a classification of paddy seed varieties by using transfer learning of deep learning with three pre-trained weights: VGG16, InceptionV3, and MobileNetV2. Over 1200 images of paddy seed are collected for custom datasets. The experimental result shows the overall accuracy of VGG16, InceptionV3, and MobileNetV2 are 80.00%, 83.33%, and 83.33% respectively. Test Loss of each model is 52.15%, 28.41%, and 61.95% sequentially. Therefore, InceptionV3 not only gives the best accuracy than others but also the least test loss. This proposed research can be a great advantage for a farmer to identify paddy varieties that aimed to reduce the adulteration of paddy varieties as well.