Problem Statement

Some activities in your facility are dispersed over broad areas. We can use shift change as an illustration. If you operate more than one shift, it is a challenge to orchestrate the handovers at shift change. It is not a simple, linear process. There may be many points where a handoff must be made. How do you optimize this parallel activity, especially when it is spread across the work space?


You installed a video system for operations with overlapping 4K cameras that view the entire work area, plus some HD cameras at localized points of interest. You record and retain the video feeds for weeks, months or more. Finally, you subscribe to a video analysis tool like Dartfish (


This system will see most of what happens in your facility. If it isn’t caught by localized cameras, an indication should be in the high definition overview images. Review the recorded video streams and find segments that show recent shift changes. Export the clips, assemble them in a timeline and look for things that are interesting or patterns that stand out.

Import the shift change clips into Dartfish and build a “tagging panel” where a viewer can flag the events of note with one click. Review the clips to add event tags and export them to a data file. The next steps depend on which patterns are interesting or important. Several promising analyses might include:

  • Measure the changeover time at different work areas and stations. Ideally, you want roughly similar changeover times. If not, some centers may be bottlenecks.
  • Look for “variation” in readiness times across the shift changes. Times at workstations should be roughly the same from one handover to the next. If there is variation, perhaps something else is distracting from the process.
  • Any other analysis that suggests itself …

The key to this approach is having a video operations system that constantly records everything. You don’t have to explicitly plan to study the shift change because the video and data are always in the video record … waiting for analysis.