Haricot beans, also known as “Boleqe” in the locale, are also known as dry beans, common beans, kidney beans, and field beans. It is a significant legume crop grown worldwide. This is an improved fabrication request for the safest and most effective way to select the most haricot beans for a specific presentation. The three basic qualities of haricot beans are color, shape, and size; nevertheless, traditional techniques based on mechanical or optical evaluation are insufficient. Inspection of flaws and quality should be developed. This study developed a method for checking haricot beans for quality and faults using machine learning techniques. In this study, we combined feature extraction using the histogram of oriented gradients (HOG) and a convolutional neural network (CNN); classification with CNN, the k-nearest neighbor technique (KNN), and a support vector machine (SVM); and finally, accuracy comparison and contrast. Using some of the features listed above, we can identify the beans. We used a scanner image from Ethiopia’s Amhara area, namely the Kombolcha Ethiopian Commodity Exchange (ECX). Our methodology included data collection via scanner images, pre- and post-processing, feature extraction, and detection of Ethiopian haricot beans. The method classifies data with 94% and 99.4% accuracy for SVM and KNN, respectively, using HOG feature extraction. Our findings indicate that our method incorporates the CNN and HOG feature extraction methods into the feature extraction process from haricot bean images. Therefore, we achieved 100% accuracy using the SVM classifier and hybrid feature extraction (CNN + HOG).