Real-time defect and freshness inspection on chicken eggs using hyperspectral imaging

Chicken eggs are a common raw material for cooking, as well as a common agricultural product. Eggs contain abundant nutrients such as oil and fat, protein, and vitamins. They are beneficial to human bodies. Residual dirt or breakage before or after the washing and selection procedure would affect the egg quality. Defect inspections for eggs were done manually in the past. As this process is labor intensive, fatigue-induced errors are likely to occur, leading to inconsistent quality. This study developed hyperspectral egg defect inspection technique (HEDIT) and hyperspectral egg freshness inspection technique (HEFIT), based on the perspectives of factory production lines and consumers. HEDIT and HEFIT combined band selections with deep neural network (DNN), one-dimensional convolutional neural network (1D-CNN), two-dimensional convolutional neural network (2D-CNN), three-dimensional convolutional neural network (3D-CNN), and MobileNet to implement real-time in-spection using hyperspectral imaging (HSI). In the experiments, 1000 fresh egg samples and 800 defective egg samples were tested. The experimental results show that the overall accuracy of our proposed method was 99% in freshness and 100% in defect inspection. Additionally, in the defect inspection, the inspection time was 31 ms, and the overall accuracy remained at 95%, maintaining a certain degree of accuracy and sorting speed. These results are beneficial to subsequent applications and the commercialization of smart egg production.