Among other grain-based foods, rice is an important and favorite food of Pakistan. Its demand has increased significantly in the recent era. The type of rice grains is very important for export and import. Therefore, it becomes necessary to distinguish rice types to avoid their fraudulent labeling. With similar aims, a dataset consisting of seven rice varieties, i.e., Basmati 2000, Chenab basmati, KSK 133, Kissan basmati, KSK 434, PK 1121 aromatic, and Punjab basmati, mostly cultivated in Pakistan have been collected and used for rice seed classification. Images of rice seed have been obtained using a self-designed setup, which consists of different seeded, i.e., 1-seeded, 5-seeded, 10-seeded, 15-seeded, and 20-seeded, images. A new model has been designed specifically for rice seed classification using convolutional neural networks, which consists of 18 layers. Standard evaluation measures based on results obtained through the proposed model have been evaluated and compared with both recent state-of-the-art published studies, and state-of-the-art CNN models, i.e., VGG-19, ResNet50, and GoogleNet (Inception-V3). Experiments proved that the proposed model achieved a perfect classification rate, i.e., 100% for each of the Pakistani grown rice seeds. This study is currently integrated with the agro-technology industry for the auto-classification of rice kernel seeds as a demo version.

在其他谷物食品中，大米是巴基斯坦的重要食品。在最近的时代，其需求已大大增加。稻米的类型对于出口和进口非常重要。因此，有必要区分大米类型以避免其欺诈性标记。出于类似的目的，已收集了一个主要在巴基斯坦种植的，由七个水稻品种组成的数据集，即Basmati 2000，Chenab basmati，KSK 133，Kissan basmati，KSK 434，PK 1121芳香族和旁遮普邦的巴斯马蒂。水稻种子的图像已使用自行设计的设置获得，该图像设置包含不同种子的图像，即1种子，5种子，10种子，15种子和20种子的图像。使用卷积神经网络专门为水稻种子分类设计了一个新模型，它由18层组成。基于通过拟议模型获得的结果的标准评估措施已得到评估，并与最近发表的最新研究以及最新的CNN模型（即VGG-19，ResNet50和GoogleNet）进行了比较（Inception-V3）。实验证明，提出的模型达到了理想的分类率，即巴基斯坦种植的每种水稻种子均为100％。这项研究目前已与农业技术行业集成在一起，用于将稻仁种子自动分类为演示版本。实验证明，提出的模型达到了理想的分类率，即巴基斯坦种植的每种水稻种子均为100％。这项研究目前已与农业技术行业集成在一起，用于将稻仁种子自动分类为演示版本。实验证明，提出的模型达到了理想的分类率，即巴基斯坦种植的每种水稻种子均为100％。这项研究目前已与农业技术行业集成在一起，用于将稻仁种子自动分类为演示版本。