Search

In recent years, the technology of deep learning has made great achievements in the field of machine learning. In this study, with the help of the transfer learning method, a kind of soft sensor is designed for the classification of iron ore tailings grade. Firstly, a sample database of froth images of flotation tailings was established. Secondly, the three most reliable models are determined after comparing the accuracy of 13 deep neural network models applied in the flotation froth image. A more accurate hybrid deep neural network model is established, with an accuracy of 97%. Finally, a software system is designed and developed, which can operate stably in the flotation plant. The experimental results show the effectiveness of the proposed hybrid deep neural network in the field of iron ore froth flotation. [ABSTRACT FROM AUTHOR]

• Morphological processing and feature value extraction of the collected dried fruit images of Z. armatum DC. • Convert RGB color model to HIS color model to obtain color characteristic values. • BP neural network model was constructed to identify and classify the dried fruit of Z. armatum DC. In order to achieve rapid and accurate grading of Zanthoxylum armatum DC. dried fruit, this study utilized image processing technology combined with BP neural networks for the dried fruit of Z. armatum identification to enhance the market value of Z. armatum and their products. The results showed that preprocessing the images with a yellow background yielded the best results, effectively distinguishing the external contours and internal images of Z. armatum. The extracted image of the dried fruit can represent information about its peel, with R, G, and B values being the largest at 69.3285, 65.6432, and 31.2561, respectively. Among the seven drying methods, the average G value was highest at 71.0560 under the conditions of blowing at 42℃, whereas it was lowest at 47.3840 under sunlight drying. Moreover, R, G, B, H, S, and I represent the input layer nodes of the BP neural network. The change in the number of hidden layers did not correlate with the classification results. The classification accuracy of the BP neural network for Z. armatum classification under all dry conditions was 79.53 %-98.04 %, indicating high reliability. The proposed method can provide a theoretical basis for future research on Z. armatum classification. A schematic diagram of color grading of green Sichuan pepper(Zanthoxylum armatum DC.)dried fruit. [Display omitted] [ABSTRACT FROM AUTHOR]

To achieve high yield of crops and to avoid pesticide abuse, different control methods have been adopted according to different degrees of disease in crop plants. To address the issue, a leaf disease grade identification method based on a convolutional neural network (CNN) was proposed. First, nonuniform illumination images were processed using an adaptive adjustment algorithm based on a two-dimensional (2-D) gamma function. Then a threshold segmentation method was used to segment diseased leaf images and thus to obtain binary images. Next, the ratio of the number of pixels in the lesion area to that in the diseased leaf area was calculated. This ratio was regarded as the classification threshold of the disease grades, and therefore was used to determine the disease grade category. In addition, use of a ResNet50-based CNN was proposed to identify disease grades, with a focal loss function replacing the standard cross entropy loss function, and with the Adam optimization method. Finally, leaf disease grade identification was performed on a database containing 10 types of disease leaf images for 8 crops, and it yielded a recognition accuracy of 95.61%. The experimental results showed that the proposed method was feasible and effective for the classification of leaf disease grades. [ABSTRACT FROM AUTHOR]