Online shopping has been advancing quickly since its popularization in recent years, people buy different kinds of products from the comfort of their houses, but sometimes these products may be returned, as they may not have looked as stylish as shown on their website, or simply because they do not fit properly on the customer’s body. According to the US National Retail Federation, the merchandise return is estimated to hit $890 billion US dollars, which is about 16% of sales (1). This encourages the companies to not only have better return policies but to innovate in ways to sell their products so that the customer is satisfied with their purchase and does not have to return it. One way of doing this is by implementing Virtual Try On’s on their virtual platforms.
A Virtual Try-On, allows customers to virtually try on the products before buying them. Which resembles the in-store fitting experience using augmented reality (AR) and artificial intelligence (AI) with machine learning (ML) algorithms. The implementation of the Virtual Try On’s is already ongoing in different industries, the most impacted the clothing and beauty industries.
In the clothing industry, one of the methods for doing so is VITON (2) which takes a selfie of a person and an image of the desired cloth and superimposes them in a way that the person’s features and pose are preserved, but the original person’s clothing changes to the one desired, matching textures and forms of the original cloth image, creating a visual representation of the desired cloth on the customer’s selfie.
In the beauty industry, one of the methods is by training a generative adversarial network GAN, like in BeautyGAN (3), so that it learns how to change the customer’s facial image in way that expresses the desired makeup on the customer’s face. In BeautyGAN, the model was trained to learn how to transfer the makeup from one face to a given one, by trial and error in which the discriminator model in the GAN was trained to classify between the originally sampled real images and the ones made by the generator. While the generator learned how to create images that kept the facial identity of the given face, but also implemented the desired makeup.
The previously explained Virtual Try On models are just examples of the efforts that have been occurring to implement a way in which a customer can engage better with a desired product shown on a website or a mobile app. As explained by McKinsey (4), customers are more excited by the experiences that augment and complement their experiences in the physical world, especially the ones that help them visualize how a product looks on them.
In conclusion, a Virtual Try-On is a promising way to showcase a product, allowing the customers to have a more engaging shopping experience, that gives them a virtual representation of how they will look, which will aid in tackling the increasing costs of handling the returns and increase the customer’s satisfaction after purchase.
References:
https://www.forbes.com/sites/shelleykohan/2024/12/08/retail-returns-surge-to-890b-how-retailers-are-adapting-in-2024/
https://openaccess.thecvf.com/content_cvpr_2018/papers/Han_VITON_An_Image-Based_CVPR_2018_paper.pdf
http://www.linliang.net/wp-content/uploads/2018/10/ACMMM2018_BeautyGAN.pdf
https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/unlocking-commerce-in-the-metaverse
