Article Preview
Top1. Indroduction
In today's digital era, the e-commerce industry is continuously expanding (Wang & Chang, 2021), and the widespread application of computer vision technology has brought revolutionary changes to e-commerce platforms (Yang & Liu, 2021). Computer vision is a technology that enables machines to understand and interpret images or videos (Sheela, 2022). Its applications are not limited to the fields of art (Manovich, 2021) and entertainment (Erdelyi, 2019); it also plays a crucial role in the business world (Soni et al., 2020). The application of computer vision technology in e-commerce provides users with a more convenient and intelligent shopping experience. Through image recognition techniques (Mehmood et al., 2019), consumers can use visual searches to find specific products without the need for text descriptions. This intuitive search method not only enhances user satisfaction but also accelerates the shopping decision-making process, leading to an increase in sales. Furthermore, by analyzing users' shopping history, browsing habits, and preferences, computer vision systems can recommend products that users might be interested in (Zhou, 2020). This personalized recommendation not only increases the likelihood of user purchases but also strengthens the bond between users and e-commerce platforms, encouraging users to visit the platform more frequently. Additionally, computer vision technology can be used to enhance the sales forecasting capabilities of e-commerce platforms (Jain & Wah, 2022). By analyzing product images and videos, the system can identify product features and predict demand trends. This precise sales forecasting helps e-commerce platforms better manage inventory, avoiding situations of surplus or stockouts, thus improving the efficiency of the supply chain and reducing inventory costs. The application of computer vision technology in e-commerce not only enhances the user experience but also strengthens the interaction between e-commerce platforms and users. Simultaneously, it improves sales forecasting and inventory management efficiency. With the continuous innovation and development of computer vision technology, it will continue to bring more opportunities and challenges to the e-commerce industry, driving the entire industry towards a more intelligent and efficient direction.
The extraction of image features from e-commerce platforms using deep learning technology (De la Comble et al., 2022) holds significant importance for predicting product sales. Deep learning models, particularly Convolutional Neural Networks (CNNs) (Liu, 2022), have the capability to automatically learn features from images without the need for manual feature extractors. Consequently, deep learning technology efficiently extracts critical features, including color, shape, texture, among others, from product images, providing more representative and enriched data for sales prediction models. Deep learning models typically exhibit strong generalization abilities when dealing with large-scale data, enabling them to adapt to various product types and market conditions. Thus, models based on deep learning technology can more accurately predict sales, extending beyond specific categories or time periods. Furthermore, product sales are influenced by a multitude of factors (Cheng et al., 2019), often characterized by complex nonlinear relationships. Deep learning models can learn intricate features and relationships within the data, better capturing the complex associations between product sales and image features, thus enhancing prediction accuracy. In addition, product images on e-commerce platforms often exhibit variations in size, angle, lighting, and other characteristics. Traditional feature extraction methods may struggle to handle this diversity. Deep learning models, on the other hand, effectively manage varying image data by using operations such as convolutions, enhancing the model's adaptability to changes in image characteristics. In summary, the extraction of image features from e-commerce platforms using deep learning technology is highly significant for product sales prediction. It not only provides more accurate and diverse feature representations but also handles complex relationships and adapts to diverse image data. It enables the development of end-to-end prediction systems, providing e-commerce businesses with a more precise and efficient tool for sales prediction, aiding in the formulation of market strategies, inventory management, and enhancing sales performance. The deep learning models commonly used in the research on e-commerce platform sales forecasting are as follows: