Foreword: This was a consulting project to assess the feasibility of a computer vision based solution to track order preparation. We were not given access to the final recommendation due to confidentiality concerns.

Buisness Challenge
Accuracy in order preparations at a fast food restaurant is of extreme importance. Having orders returned by customers has a negative impact not only on the customers but also on the kitchen’s flow, generating delays.

This fast food restaurant chain has installed video cameras in their kitchens and were investigating whether they would be able to find and track orders in the video streams.

Supplier Solution
The supplier investigated the feasibility of developing an AI Video Analytics solution to track orders and detect their ingredients while being prepared.

Outcome
Multi-label object detection and object tracking under heavy occlusions were the main aspects investigated.

A labelling guideline and a first version of a multi-label object detection and tracking system were developed - a proof of concept (a more affordable, cut-back and sandboxed version to demonstrate the feasibility).

Recommendations based on the complexity of the problem, the findings during the project and previous experience in CV solutions were delivered.