Find us

Automation of the process on the shopping platform

Story

The goal of KidiHub was to create an automated platform that would enable:

  • Automatic descriptions of the children’s clothes being exchanged.
  • Precise classification of clothes according to type, color, pattern, style, and size.
  • Facilitation of quick and seamless exchanging and acquiring of used children’s clothes for users.
  • 24/7 availability.

Problem

Our client was spending a lot of time preparing, verifying the quality, and pricing children’s clothes.

They were looking for technology to optimize the entire process.

We thoroughly analyzed the process and proposed a solution that classifies the clothing into a specific category, measures it, and assigns predefined attributes

Solution

We applied various types of models – from classical computer vision, through simple heuristics, to the use of deep convolutional networks.

Thanks to the implementation of this technology, the system:

  • Recognizes the type of clothing (e.g., skirt, dress, blouse).
  • Identifies the dominant color and pattern (plain, patterned, with text).
  • Recognizes the style (vintage, casual, elegant, sporty).
  • Precisely determines dimensions: a special algorithm measures the dimensions of the clothing based on photos taken on a standard measuring table, where a ruler serves as a reference system.
  • Assigns the appropriate set of attributes: tailored to the specifics of the given type of clothing (e.g., different attributes for a blouse, different for gloves).

Thanks to the use of artificial intelligence, the entire process has been optimized.

The automation of clothing description and classification has significantly reduced the time needed to prepare clothes for circulation, resulting in increased operational efficiency. Accurate measurements and detailed descriptions provide users with complete information about the exchanged and acquired clothes, enhancing service quality. With faster clothing turnover, the platform offers a greater availability of diverse products, supporting parents in quickly and easily exchanging and acquiring children’s clothing, often for a symbolic fee or in exchange.

The innovative technology utilizing advanced generative models and special algorithms for describing and measuring clothes provides the platform with a competitive advantage. The integration of various functions into one platform creates a comprehensive solution that can be easily scaled to other markets, and also allows for the introduction of additional product categories

Technology

We applied various types of models – from classical computer vision, through simple heuristics, to the use of deep convolutional networks.