OKHome / Projects / Code and cap on bottles
Case study · Dairy · Bottle
Bottles without a code: from complaints to reading every pack
A dairy plant was receiving complaints about bottles without a code. Its existing system only checked whether something was printed or not. After several months of trials on the line, we replaced it with three tools that read the code, check the cap and measure how it is fitted.
- Complaints about missing codes
- Solved
- Tools per bottle
- 3
- Cap
- Deep learning
The starting point
Seeing that something is printed is not enough
A presence or absence inspection accepts a blurred, incomplete or half-printed code as good. The bottles went out, and the problem surfaced at the customer.
That is why the answer was not to adjust the previous system, but to change what is checked. We ran trials on the line itself, over several months, until we found the combination of tools that solved the problem.
The solution
Three tools on every bottle
- OCR of the code, with logging. What is printed on every bottle is read and logged; if it cannot be read, the bottle is removed from the line.
- Cap presence by deep learning. It learns from examples of good and bad bottles.
- Cap placement angle. It detects capping deviations before a badly fitted cap causes product loss.
The result
Every bottle, with its code read and its cap properly fitted
The complaints about missing codes were solved, and the cap angle measurement warns of capper deviations before product is lost.
More on these checks: date and batch, caps and closures and deep learning in inspection.
Free remote feasibility test
Does every bottle leave with its code?
Send us photos or videos of your good and bad bottles. We will tell you what can be checked on the line.