MLOps for Auto Attribution of Images Using Computer Vision

About the Customer & Challenges Faced:

This is a major American designer and marketer of children’s apparel. The use of a line assortment application for mapping product attributes to facilitate planning. This process is manual-intensive and takes a lot of time. On average there are 130 attributes per product and around 4 million attributes are being managed per year. The sales and marketing teams needed a new, better way to describe the product. They wanted to leverage artificial intelligence for recognizing images of children’s clothing and predict labels with respect to certain attributes such as sleeve length, leg length, and patterns to reduce manual intervention. The situation demanded the need for an integrated platform for data scientists equipped with accelerators and tools to manage the entire AI application development life-cycle.

Solution and Approach:

Based on the above problem statements, collaborated with the client to build image recognition and auto attribution solution using the platform.

  • Developed a VGG-19 ConvNetDNN model that takes an image of the piece of clothing as input and outputs a set of labels for the image.
  • Used transfer learning to dramatically improve development time and model performance by utilizing a pre-established model architecture and weights
  • The model architecture and weights are loaded from the TensorFlow. Keras applications module
  • Attribute examples:
    a. Primary leg length: leg long, leg short, or neither
    b. Primary sleeve length: sleeve length, sleeve short, or neither
    c. Primary leg type: cinched, jogger, legging, short, footed, jegging, skirt, or neither
    d. Primary sleeve type: ruffle sleeve, raglan, or neither
    e. Primary character: approximately 10 different designs
  • The solution scans the uploaded image, identifies relevant components, assigns relevant attributes and populates the attributes automatically on the line assortment application


  • By using, one can leverage high-end data connectivity, efficient data versioning, perform exploratory data analysis and generate inferences using an intuitive process and through an industry-standardized manner.
  • The unique, containerized platform-centric approach offered by can be used to employ required infrastructure, deploy rapidly to multiple high-availability environments while aligning with best-in-class DevSecOps practices.
  • also brings in-depth QA-QC testing and logging frameworks, synchronous and asynchronous monitoring, and performance tracking ability.
  • also has SSO (single-sign-on) for various in-built tools and subsystems that make the platform access seamless throughout.​
  • In a nutshell, all the above features in a single plate under the same hood make an unbeatable AI Ops framework.

Have Any Questions?

Need more information about the platform?