Waste management using an automatic sorting system for carrot fruit based on image processing technique and improved deep neural networks

In this study, we address the problem of classification carrot fruit in order to manage and control their waste using improved deep neural networks. work, perform a study show that convolutional networks are straightforward approach solve problem. Additionally, improve network (CNN) based on learning pooling function by combining average max pooling. We experimentally merging operation used increases accuracy compared other methods. For purpose, images 878 samples various shapes (regular irregular) were taken after preprocessing operation, they classified CNN. To compare method with methods, image features extracted Histograms Oriented Gradients (HOG) Local Binary Pattern (LBP) methods Multi-Layer Perceptron (MLP), Gradient Boosting Tree (GBT), K-Nearest Neighbors (KNN) algorithms. Finally, proposed CNN algorithm, was The results showed 99.43% for grading through configuring Batch Normalization (BN)-CNN mixed Therefore, can be effective increasing marketability, controlling improving traditional fruit.