Applications of convolutional neural networks for intelligent waste identification and recycling: A review

With the implementations of "Zero Waste" and Industry 4.0, the rapidly increasing applications of artificial intelligence in waste management have generated a large amount of image data, with concomitant improvements in the analysis methods. As advanced image analysis approaches, convolutional neural networks (CNNs) have become indispensable tools for finding hidden patterns in visual features. Over the last few years, CNNs have been progressively applied to a wide variety of intelligent waste identification and recycling (IWIR). However, CNNs are still new to environmental researchers, and current studies on IWIR are hard to summarize due to the lack of benchmarks and widely accepted standards for the datasets and models used. Therefore, the aim of this review was to examine CNN approaches and their applications in IWIRs. First, some essential knowledge of CNNs was introduced. Then, the various open-source datasets and advanced CNN models used in IWIR were outlined, with insights into the three main tasks: classification, object detection, and segmentation. Then, three key fields of CNN applications in IWIR, i.e., recyclable material identification, trash pollution detection, and solid waste classification, were summarized. Finally, the challenges and limitations of the current applications were discussed to elucidate the future prospects of CNNs in this field.