During the past 5 years, the number of online footwear retailers has been steadily increasing at a 5.7% rate per year. Sales for this online industry has also gone upward. And this trend is not likely to change, with IBISWorld estimating that the industry’s revenue will keep rising at an annualized rate of 6.3% until 2023.
With Internet taking over the retail market, online sales have become one of the main revenue channels, accounting last year for around 22% of total footwear spends. Shoppers buy footwear less frequently than clothing (9.9% of shoppers says they buy shoes at least once a month in front of 15.7% that buys clothing monthly). However, in U.S., online shoe shops are still performing better than other segments of retail.
Great opportunity for footwear e-tailers
In 21st century, consumers are far more knowledgeable and have access to all kinds of information in a matter of seconds. Shoemakers are faced with this change in customer behavior. It makes competition more stiff than ever and requires a huge ability to constantly adapt offerings to consumer preferences. These are key to be able to influence purchasing decisions.
In traditional brick-and-mortar footwear stores, customers might not be able to see special designs or compare prices with the same comfort as they can do it online. Browsing 10 different online stores requires much less time and effort than entering 10 physical stores and walking around trying to find the perfect pair of shoes. Online footwear marketplaces provide customers a better product range overview. They also enable different filters that help shoppers narrow the query to look only at relevant items in a blink of an eye.
The growing challenge of product annotation
Being in a fast-paced, growing industry carries several operational challenges. Online footwear catalogs are constantly being updated with new products coming up everyday.
A massive collection of shoes needs intensive human labor to manually annotate every single product. It takes time to define whether a shoe belongs to sneakers or a formal shoe category, and even more effort to accurately describe the type of shoe closing, material or shape. Automatic product image tagging systems can do so much faster and down to the most fine-grained details.
Automated tagging for images of shoes
Artificial Intelligence systems trained with thousands of images are able to recognize images of shoes and extract semantic attributes while also classifying them into the corresponding categories. This way, it saves time and money to footwear e-tailers, and helps optimize their workflow and product cycle in order to speed time-to-market up.
Catchoom’s deep learning models perform these actions really fast and reduce the need for manual labor. It can classify footwear into different categories, ranging from sport shoes to pumps or sandals, and into different subcategories. For instance, sport shoes can be divided into tennis, running or golf subcategories, to mention some examples. And even some of these subcategories have a few secondary subcategories themselves.
Moreover, the solution also assigns different attributes by recognizing features in the image. It can tell whether the shoe has stiletto, cone, block heels or none, for example. It also detects the type of shoe closing (straps, velcro, buckle laces, etc.), as well as the material (metal, synthetic, rubber, etc.) and shape (flat, rugged, serrated etc.) of the sole.
With 150+ categories (i.e., office pants) and 40+ attributes (i.e. length, pattern…) with its +400 different values, Catchoom’s fashion-specific AI solution is one of the finest-grained solutions to automatically tag fashion product images.