Deep Learning for Retail Product Recognition: Challenges and Techniques

Abstract

Taking time to identify expected products and waiting for the checkout in a retail store are common scenes we all encounter in our daily lives. The realization of automatic product recognition has great significance for both economic and social progress because it is more reliable than manual operation and time-saving. Product recognition via images is a challenging task in the field of computer vision. It receives increasing consideration due to the great application prospect, such as automatic checkout, stock tracking, planogram compliance, and visually impaired assistance. In recent years, deep learning enjoys a flourishing evolution with tremendous achievements in image classification and object detection. This article aims to present a comprehensive literature review of recent research on deep learning-based retail product recognition. More specifically, this paper reviews the key challenges of deep learning for retail product recognition and discusses potential techniques that can be helpful for the research of the topic. Next, we provide the details of public datasets which could be used for deep learning. Finally, we conclude the current progress and point new perspectives to the research of related fields.

Deep Learning in Barcode Recognition: A Systematic Literature Review

“…Researchers should also balance the tradeoff between model generalization and detection accuracy for DL-based barcode recognition. The trade-off balance execution may be realized by enabling proper image transformation operations or adopting augmentation-based generative models, as suggested in [12]. We assert that each augmentation technique has specific strengths that can adaptively improve model generalization while retaining as much accuracy as possible.…”

“…Rather than that the practical applications of DL technologies, especially CNNs have been proved to be useful for analyzing data in a form of graph and manifold, namely geometric DL [31]. These distinctive advantages, as well as the rapid progression of DL alongside technological advancements (i.e., computing power and visual performance, devices' image resolution, and the improved cost-VOLUME XX, 2017 effectiveness of both hardware and software) [12], [30], have accelerated the application of DL across several domains, such as medical, SCM, and manufacturing industries.…”

“…Likewise, one of the major challenges of barcode detection based on the DL approach is the data used in the learning process [14]. [12], [71], [72]. The datasets that fall within this class are considered secondary sources or reusable data.…”

“…Similar to the abovementioned review studies of DL in CV, the number of literature review studies of D-CNNs and CV in topics relating to daily life or manufacturing tasks (e.g., barcode detection) is also limited. Only four related review articles examined different topics or scopes, including retail product recognition [12], material degradation analysis [13], collision avoidance for unmanned aerial vehicles (UAV) [14], and plant disease detection [11].…”

Classification of Retail Products: From Probabilistic Ranking to Neural Networks

“…Deep learning technologies are also used in this field (retail) to enhance the information analysis processes. In fact, [24] offers a survey of the most recent works in the field of computer vision applied to automatic product recognition combined with deep learning techniques. Other interesting analysis that supplements the previous one is the work in [25].…”