On the other hand, when deep learning (DL) approaches were used such as recurrent convolutional neural network (RCNN) or you only look once (YOLO), high accuracies could be achieved. In addition, the visual similarity among the different products of the same brand can lead to misclassification. When traditional image processing methods have been used, such as histogram of oriented gradient (HoG) for feature extraction and support vector machine (SVM) for a classifier, they have limited performance even if on the large datasets and their performances are hard to be increased. When all these previous works are examined, there are pros and cons of these approaches. Some of them applied traditional image processing techniques to detect the presence and absence of the product, such as, and some of them used deep learning approaches for object detection on the shelves, such as. One of them proposes radio frequency identification (RFID) tagging to monitor product quantity on the shelves, but this approach is not cost-effective to implement the technology and integrate it into existing systems. These studies consider the subject from different perspectives. There are several studies to automate monitoring OSA. This approach is not effective and sustainable since it continuously requires human effort. OSA is checked by employees manually at most of the grocery stores. Current inventory systems cannot understand the number of products on the shelves. These products, available in stock, might not be on the shelves. The remaining products can be checked using an inventory management system, but it only shows the number of products in the stock. For this reason, OSA has a significant effect on business profit. Besides, another study showed that the “out of the stocks” rate is about 8% in the United States and Europe. According to the research reported by Corsten and Gruen, when a product is not available on the designed shelf space, 31% of consumers buy the product from a different store, 26% of them buy a different brand, 19% of them buy a different size of the same brand, 15% of them buy the same product at a later time, and 9% of them buy nothing. When a product that a shopper looks for is not available on its designed shelf, also known as “out-of-stock” (OOS), this causes a negative impact on customer behaviors in the future. Providing high OSA is a key factor to increase profits. One of them is the monitoring on-shelf availability (OSA) in grocery stores. Machine learning techniques have been applied to different areas in the retail sector. The experimental results show that the proposed approach outperforms the existing approaches (RetinaNet and YOLOv3) in terms of accuracy. In the experimental studies, the effectiveness of the proposed SOSA method was verified on image datasets, with different ratios of labeled samples varying from 20% to 80%. It presents a new software application, called SOSA XAI, with its capabilities and advantages. Furthermore, this paper provides the first demonstration of explainable artificial intelligence (XAI) on OSA. Moreover, it is the first time that “You Only Look Once” (YOLOv4) deep learning architecture is used to monitor OSA. To tackle the annotation problem, this paper proposes a new method that combines two concepts “semi-supervised learning” and “on-shelf availability” (SOSA) for the first time. However, the largest and well-known computer vision datasets do not provide annotation for store products, and therefore, a huge effort is needed to manually label products on images. Recently, there has been growing interest in computer vision approaches to monitor OSA. Providing high on-shelf availability (OSA) is a key factor to increase profits in grocery stores.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |