Large amounts of agricultural wastes are generated in agricultural production, and composting this waste is one of the best ways to recycle resources. Compost maturity is an important criterion for measuring the quality of compost-products. Biochemical tests are conventional methods to evaluate compost maturity, but they are time consuming and difficult to perform. Therefore, convolutional neural networks (CNNs) were introduced to realize fast evaluation of compost maturity by analyzing images of different composting stages. Images of 3 different composting materials were collected to build 4 data sets, which included nearly 30,000 images, and a series of experiments were performed on them. The accuracy of proposed method was 99.7%, 99.4%, 99.7% and 99.5% on the 4 test sets, respectively. Experimental results demonstrate that the proposed CNN-based prediction model produces state of the art results and can be used to predict compost maturity during the composting process.
Keywords: Agricultural waste; Compost maturity; Convolutional neural networks; Image analysis.
Copyright © 2019 Elsevier Ltd. All rights reserved.