Food container seal inspection is a common application
Fruit pieces are deposited into a pre formed cup and a flexible seal is placed which is thermo bonded in place. Fruit pieces, especially pineapple, can become caught in the seal area and cause leakage and spoilage of the product.
The objective is to inspect all seals and remove defective product on the production line before packaging.
This facility had 12 lines processing fruit cups. Each line ran at 6 cups per second.
The cups were all the same size but a variety of pre printed color film was used as the sealing web. This created some issues for the image processing software.
Defective cups needed to ejected immediately after the inspection station.
The system needed to record the total defects detected per batch as well as monitor a running average in order to detect faults in the sealing process before it became a huge issue.
The image processing functions needed to be quite fast. If a reject was detected, the part needed to be ejected at the adjacent reject station. The process time per part image was only a few milli seconds.
Deep learning technology was used to account for the random nature of the defects.
The system as deployed did not require operator intervention on a routine basis.
The last defect images were displayed and could be logged if required for defect tracking purposes.
In addition the system could "track" the running defect average and warn of impending problems before they became a serious process issue.
Process speed was adequate. In this case each part took 0.036 sec to process.
This means that the image processing system was adequate for the purpose. Also note that the system could apply a "confidence" value to confirm the level of certainty. This feature could be used to futher refine system performance.
To find out how Indyn can help with your particular application, just click the link belowdiscuss