The Potential Application of Innovative Methods in Neural Networks for Surface Crack Recognition of Unshelled Hazelnut DOI Open Access
Alireza Shojaeian, Hossein Bagherpour, Reza Bagherpour

и другие.

Journal of Food Processing and Preservation, Год журнала: 2023, Номер 2023, С. 1 - 9

Опубликована: Ноя. 8, 2023

In some countries, most hazelnuts are cracked using semi-industrial or hand-crafted machines and marketed as open-shell. the process of hazelnut cracking, because different sizes shapes hazelnuts, many leave cracking machine in form a closed-shell. The presence closed-shell reduces marketability product. Therefore, after operation, separation closed-shells from whole has largely been conducted by visual inspection, which is time-consuming, labor-intensive, lacks accuracy. So, purpose this study was to use deep convolutional neural network (DCNN) algorithm classify into two classes: open-shell hazelnuts. To compare proposed method with pretrained DCNN models, three models including ResNet-50, Inception-V3, VGG-19 were investigated. results model (accuracy 98% F 1 -score 96.8) showed that good capability predicting classes. Compared small size simple architecture model, can be substitute for complex large such Inception-V3. Overall, indicate crack on surface successfully detected automatically, high potential facilitate development sorter based crack.

Язык: Английский

The Potential Application of Innovative Methods in Neural Networks for Surface Crack Recognition of Unshelled Hazelnut DOI Open Access
Alireza Shojaeian, Hossein Bagherpour, Reza Bagherpour

и другие.

Journal of Food Processing and Preservation, Год журнала: 2023, Номер 2023, С. 1 - 9

Опубликована: Ноя. 8, 2023

In some countries, most hazelnuts are cracked using semi-industrial or hand-crafted machines and marketed as open-shell. the process of hazelnut cracking, because different sizes shapes hazelnuts, many leave cracking machine in form a closed-shell. The presence closed-shell reduces marketability product. Therefore, after operation, separation closed-shells from whole has largely been conducted by visual inspection, which is time-consuming, labor-intensive, lacks accuracy. So, purpose this study was to use deep convolutional neural network (DCNN) algorithm classify into two classes: open-shell hazelnuts. To compare proposed method with pretrained DCNN models, three models including ResNet-50, Inception-V3, VGG-19 were investigated. results model (accuracy 98% F 1 -score 96.8) showed that good capability predicting classes. Compared small size simple architecture model, can be substitute for complex large such Inception-V3. Overall, indicate crack on surface successfully detected automatically, high potential facilitate development sorter based crack.

Язык: Английский

Процитировано

0