The challenge
On the bacon line, each bacon belly was manually measured by an employee using a ruler: first the width (from the shoulder side to the ham end), then the height at the thickest point. Based on the results, the product was manually placed into one of five classification categories.
This process placed high demands on the employee's concentration and was susceptible to measurement errors, fatigue, and inconsistency between employees. On a continuously running production line, this led to undesirable variation in classification results, with the risk of incorrect product allocation to retailers. Furthermore, the manual process did not yield measurement data—making it impossible to evaluate or adjust classification thresholds based on facts. The question was not only how to automate the measurement but also how to use the obtained data to structurally maximize yield.
The solution
Vision Partners developed and delivered mobile inspection machines that measure and classify the height and width of each bacon belly in-line, fully automatically. The system is made entirely of stainless steel and complies with IP67 classification, making it suitable for use in the wet production environment of the meat processing industry.
The core challenge was accurately measuring an irregularly shaped, wet piece of meat that is moving on a conveyor belt. Vision Partners solved this by combining two industrial 3D cameras based on Time of Flight (ToF) technology. These cameras measure the distance to the product via the travel time of light, thus building a 3D image of both sides of the moving product. The combined point cloud forms the basis for accurate dimension determination.
The system consists of:
- 2x IP66 Industrial 3D ToF Cameras: generate a combined point cloud of the bacon belly for accurate height and width measurements
- 3D vision software calculates product features (width and height) of the bacon belly from the point cloud; the classification criteria are adjustable via a configuration table
- Integrated laser line: projects a visible line onto the conveyor belt so the employee always places the product in the correct position
- 2 trigger sensors: does the measurement start automatically upon detection of the product, regardless of the conveyor belt direction
- 5 signal lights (numbered 1–5): show the classification result directly to the employee, visible until the next bacon is classified
- Touchscreen HMI: for product selection, batch number entry, and real-time display of classification results
- Machine Vision processor (Windows/Intel, Microsoft Azure Certified for IoT): provides the basis for future cloud integration and daily CSV reporting on a shared network
The result
The inspection machine measures every bacon belly fully automatically with high accuracy and consistency that cannot be achieved with manual measurement. The employee only needs to place the product on the belt; the classification result appears immediately via the signal lights. Human measurement errors and inter-rater variability are thus eliminated.
Because the batch number can be entered via the touchscreen, the measurement data can be linked to the cutting yields per batch. This opens up a completely new possibility: an internal data scientist uses the collected measurement data to analyze at which classification limits the yield of the bacon line is maximized. Where the limits were previously determined by experience and intuition, they are now continuously substantiated and adjusted based on objective measurement data. This makes classification not only more consistent but also economically more optimal.
Vision Partners delivered the solution turn-key on-site, including a feasibility study, development and construction of the inspection machine, installation and commissioning of the classification table in close cooperation with the client, and training of the employees.
