Foreign object detection using computer vision for a leading QSR
To solve for
- Approximately 2.5 to 3 lakh dough balls are prepared daily at one of the commissaries and shipped to
restaurants. There are 11 such commissaries and 1800+ stores. Plastic trays are used to carry the dough balls. - Numerous complaints received from restaurants. Issues include trays not being clean, incorrect counts of
dough balls, incorrect sizes, and dough balls containing black particles, plastic, or other foreign materials. - Trays with incorrect counts lead to shortages at the stores. If the store team does not notice the defects, then
defective pizzas is delivered to customers.
Solution
- Built a hardware & software computer vision system that is placed over the conveyor belt that can visually inspects the dough balls in the tray moving at a very high speed.
- Checks doughball size, count, sticking together, dough residue, black particles, nuts, threads, etc.
- For operator, defected tray image is shown on big screen along with alarm.
Benefits
- Defects reduced from approx. 1600-1800 per SCC to around 700-800.
- Training & Quality team remotely audit the dough ball without going to SCC.
- Store complaints reduced by 30%-40%.
- Due to each tray images on website, team can perform defect analysis.
- Tray tracking is now possible for which RFID expensive solution was planned.
