X-ray DR Images Dataset of Camellia Seeds and Chestnuts

his dataset contains preprocessed X-ray DR images of Camellia oleifera seeds and chestnuts for non-destructive multi-class defect classification. The images were further enhanced using techniques such as rotation, flipping, contrast adjustment, and brightness normalization. A total of 2,488 Camellia seed images and 2,844 chestnut images were collected. The Camellia seed dataset is categorized into four classes: healthy, shriveled, hollow (cavity), and insect-damaged samples. Similarly, the chestnut dataset includes four classes: healthy, double-kernel, moldy, and insect-damaged samples. This dataset was used to train and evaluate convolutional neural networks (VGG16, ResNet18, DenseNet121) for multi-class classification of internal defects in nut fruits. These images support research on non-invasive quality inspection, defect detection, and automated grading systems for agricultural products using X-ray imaging.