Precision Measurement for Industry 4.0 Standards towards Solid Waste Classification through Enhanced Imaging Sensors and Deep Learning Model

Abstract

Achievement of precision measurement is highly desired in a current industrial revolution where a significant increase in living standards increased municipal solid waste. The current industry 4.0 standards require accurate and efficient edge computing sensors towards solid waste classification. Thus, if waste is not managed properly, it would bring about an adverse impact on health, the economy, and the global environment. All stakeholders need to realize their roles and responsibilities for solid waste generation and recycling. To ensure recycling can be successful, the waste should be correctly and efficiently separated. The performance of edge computing devices is directly proportional to computational complexity in the context of nonorganic waste classification. Existing research on waste classification was done using CNN architecture, e.g., AlexNet, which contains about 62,378,344 parameters, and over 729 million floating operations (FLOPs) are required to classify a single image. As a result, it is too heavy and not suitable for computing applications that require inexpensive computational complexities. This research proposes an enhanced lightweight deep learning model for solid waste classification developed using MobileNetV2, efficient for lightweight applications including edge computing devices and other mobile applications. The proposed model outperforms the existing similar models achieving an accuracy of 82.48% and 83.46% with Softmax and support vector machine (SVM) classifiers, respectively. Although MobileNetV2 may provide a lower accuracy if compared to CNN architecture which is larger and heavier, the accuracy is still comparable, and it is more practical for edge computing devices and mobile applications.

Prediction of stamping parameters for imitation π-shaped lithium battery shells by building variable weight and threshold pelican-BP neural networks

“…In 1990, Picard et al 27 found that the division ratio of the training set, validation set, and training set, although not mandatory but has a certain impact on the evaluation performance, as can be seen from a large number of training studies on neural networks in recent years, most researchers have used a ratio of 80% for the training set and 20% for the rest. 30,31 For BP-ANN and its derivative models, this paper uses the ratio of the training set, validation set, and test set 8:1:1 to divide them, and for other neural network models, this paper uses the ratio of the training set to test set 8:2 to divide them.…”

“…From researchers’ studies, 30,31 it can be seen that the number of layers of the hidden layer structure of the neural network affects the prediction accuracy of the overall model as well as the computing time, especially for the prediction of the processing parameters of a very complex mechanical structure such as a pure electric vehicle power pack, it is more necessary to increase the number of layers of the structure of the neural network to improve the accuracy of the prediction, but too many layers will increase the difficulty of the calculation. This paper set up a single input layer with an output layer and increase the single hidden layer in order according to a sequence of equal differences with a tolerance of 1.…”

Education and awareness: keys to solid waste reduction

“…To ensure the success of recycling, it is essential to separate and manage waste efficiently and properly. Computational complexity in the classification of non-organic waste is directly related to the performance of peripheral computing devices [11]. Existing research in waste classification has been carried out using CNN architectures, such as AlexNet, which contains a large number of parameters and requires a great many operations to classify a single image.…”

Dry Waste Classification Using Quadratic Support Vector Machine for Intelligent Waste Management System

“…When the class allocation is skewed, there is a drawback in the fundamental "majority voting" classification [18]. That is, examples of a more regular class tend to dominate the new example prediction because they are common among k neighbors, because they are numerous [19].…”