Introduction

Picture this, how do you find the objects you need for example your lost keys? As research and our daily lives show, humans rely heavily on visual cues. From forming social interactions to making decisions in the absence of the entire picture.

Clearly, visual stimuli play a major role in our lives and recently, there have been impressive advances in the implementation of artificial vision in form of AI generated videos. Open AI’s Sora is one such example that takes user prompts and returns a video from the prompt. In basic terms, Sora begins with random noise, forms images from this noise, and combines these images into a video, frame by frame.

One area that has huge potential for utilizing the power of our visual processing system, is online shopping, which at the time of writing still heavily relies on text based searches.

Why should we shift from text based search?

Historically, following the development of databases, most search implementations have focused heavily on text. How this works is that specific products are placed in categories or tagged. The user can then search for an item by typing some text for example “a blue dress”. The search algorithm might work like: find dresses -> find dresses tagged blue.

Alternatively, the text could be tokenized into [“a”, “blue”, “dress”] and with some processing, ‘filler words’ like ‘a’ a removed leaving us with only [“blue”, “dress”] and returning items that have these tokens.

As one might guess, this has the disadvantage of being non-specific. The catalog could have 1000 blue dresses and the best product for the user could be hidden deep in the last results. In addition, a user then has to sift through lots of data to get to their actual products. This also opens up the database to lots of duplicity in terms of items.

What is the future of online shopping and e-commerce in the age of AI?

What if instead of text based searches, users could find the products they need the natural way, visually. As I build V5 Vision, I envision a future of online shopping and e-commerce that allows us to find the products we need in a more visual way.

While giants like Google have these features, they are very limited and suffer from the same issues as text based searches - too many results and a lack of personalization.

At V5 Vision, we envision a future where AI Visual Product Search is the standard for finding products online. By using the latest AI and computer vision algorithms, retailers and e-commerce businesses can allow users to find the exact products they need with minimal effort and reduced mindless scrolling. In addition, using an AI Visual Product Search by using images to search for products can lead to better personalized recommendations.

Another key feature of an AI Visual Product Search is the ability to quickly filter duplicate listings, further reducing clutter.

Conclusion

This was a short blog to outline the benefits of an image based search compared to the defacto text based search. Agree or disagree? Let me know in the comments below.