Google’s try-on feature: Ushering in inclusive online shopping

Imagine stepping into a virtual fitting room where you can try on clothes from the comfort of your own home. No more awkwardly standing in front of a mirror in a department store’s dressing room, trying to figure out if the dress in your hands is a flattering fit or not.

That’s the future Google aims to create with its latest innovation. The tech giant has recently unveiled an exciting new shopping feature that harnesses the power of artificial intelligence (AI) to revolutionize the way we shop online.

The feature breakdown: Virtual try-on

This new feature on Google Shopping allows you to virtually “try on” clothing, specifically women’s tops, for now. By utilizing images of real models across a diverse range of sizes from XXS to 3XL and varying skin colors, Google can virtually “dress” the models in AI-generated versions of clothes from hundreds of brands.

So, if you’re a fan of Everlane, Anthropologie, or H&M, you can see how these brands’ clothes might drape on a model that closely resembles you. When you find that model, you can save them to be your default model, making future online shopping experiences even more personalized.

The AI engine: Powered by Diffusion

Under the hood of this virtual try-on feature is Google’s generative AI model that uses a technique called diffusion. If diffusion sounds complex, it’s actually based on a fairly straightforward concept. Diffusion is the process of gradually adding extra pixels, or “noise,” to an image until it becomes unrecognizable, then removing the noise until the original image is reconstructed in perfect quality.

Google’s model adds a twist to the traditional diffusion process. Instead of using text as input during diffusion, like other models do, it uses a pair of images—one of a garment and another of a person. Each image is processed by its own neural network and shares information in a process called “cross-attention,” resulting in a photorealistic image of the person wearing the garment. This blend of image-based diffusion and cross-attention is what powers this new AI model, making the virtual try-on feature more realistic.

The training ground: Google’s shopping graph

To create a robust, versatile, and highly accurate model, Google tapped into its Shopping Graph, arguably the world’s most comprehensive dataset of products, sellers, brands, reviews, and inventory.

The model was trained using countless pairs of images, each including a person wearing a garment in two different poses. This way, the AI learns to match the shape of the garment in one pose with the person in the other pose, and vice versa, until it can generate realistic images of the garment on that person from all angles.

The benefits: Confidence and inclusivity in online shopping

This innovation is a major leap forward for both retailers and consumers. For one, it brings inclusivity to the forefront of online shopping. With models of varying sizes and skin tones, shoppers can now see clothes on a body type that looks more like their own. This development caters to the 42% of online shoppers who’ve reported not feeling represented by images of models.

More than that, it injects a sense of confidence into the online shopping experience. This technology aims to reduce the frustration that 59% of shoppers experience when an item looks different on them than expected. As a result, consumers can make more informed decisions about their purchases.

The caveats: It’s a work in progress

As promising as this new feature is, it’s still in its early stages. For now, it’s only available for women’s tops and only within Google Shopping. Moreover, the current version includes images for all sizes, even if a brand doesn’t offer a particular size.

While this makes the shopping experience more inclusive, it can lead to disappointment if a customer finds a garment they like, only to learn it’s not available in their size.

The future of online shopping

So, what does all this mean for the future of online shopping? Well, it signifies a shift towards a more personalized and inclusive shopping experience, one that harnesses AI to bridge the gap between brick-and-mortar stores and online platforms.

In the years to come, we might even see the application of this technology expand beyond clothing to include makeup, shoes, accessories, and more. And who knows, we may soon find ourselves living in a world where our online shopping experience is so seamless, we’ll forget what it was like to shop without the help of AI.

In conclusion, Google’s AI-powered virtual try-on feature might just be the beginning of a new era in online shopping. The retail industry has long been yearning for a solution to the challenge of fit and representation in online shopping, and with the power of AI, Google may have found an answer.