Kadans

From manufacturing video to standard time

Kadans is being developed to automatically break manual work into work elements, recognize the motions involved and calculate standard time. The first application is sewing operations in apparel manufacturing.

Analyzing a manual operation in detail takes time and specialist knowledge. Identifying the work elements and motions consistently is essential to a repeatable standard-time calculation.

Kadans aims to automate that path from video to work analysis to standard time, starting with sewing operations.

The product being developed follows three steps:

  1. Manufacturing video. Use an ordinary, single-camera recording of a manual operation.
  2. Work elements and motion analysis. Automatically separate the operation into work elements and recognize the motions involved.
  3. Standard time. Use the resulting work analysis to calculate standard time, rather than treating the elapsed time in the recording as the standard.

Kadans is in research and prototype development. The foundations currently include:

  • An initial video dataset of sewing operations.
  • A prototype pipeline that combines observations of hands and motion with visual interpretation of the work being performed.
  • An annotation tool for marking reference start and end points of work elements in training data.

Current work focuses on preparing and validating training examples and developing a model to detect work-element boundaries. Annotation supports model development; the intended product performs the analysis automatically.

The complete video-to-standard-time workflow is not yet validated, and there is no commercial deployment.

This is an early-stage, bootstrapped research and development project led by Özge Özdeş through a sole proprietorship.

For research and business enquiries.