Defect Detection and Classification of Soybean Using Convolutional Neural Network

One of the aims of this paper is to construct an operational prototype that conducts an image-processing system that utilizes the defect detection and classification of soybeans by using a convolutional neural network algorithm. The camera used for capturing and processing the soybean images is the Raspberry Pi Camera module. The researcher has created a system capable of detecting and classifying the four different defect types of soybeans and good soybeans. Based on the findings, the prototype could correctly detect and classify whether the soybeans were defective or good; however, factors such as colors and shapes could have a minor impact on the model's accuracy. Overall, the incorrect detection and classification by the prototype of deformed soybeans for damaged soybeans will fall into an unhealthy class.