Celik, YukselBasaran, ErdalDilay, Yusuf2024-09-292024-09-2920221863-17031863-1711https://doi.org/10.1007/s11760-021-02094-yhttps://hdl.handle.net/20.500.14619/4135Convolution neural network (CNN) is a deep learning technique widely used in object identification and classification. One of the objects that are identified and classified is grain products. We proposed a hybrid CNN model to identify the dataset obtained from 41 different durum wheat grains in the present study. A new deep feature set was created in the proposed model by combining Logits and Pool10 feature layers of the CNN models MobileNetV2 and SqueezeNet. This new feature set has been classified into the support vector machines (SVM) input. As a result of the experimental tests performed with the proposed hybrid model on the durum wheat data set, an accuracy rate of 91.89% was obtained. In addition, within the scope of this study, a unique durum wheat data set was publicly presented to researchers and added to the literature.eninfo:eu-repo/semantics/closedAccessConvolution neural network (CNN)MobileNetV2SqueezeNetDeep featuresSupport vector machines (SVM)Article10.1007/s11760-021-02094-y2-s2.0-8512306562311424Q2113516WOS:000742833500001Q3