Process Management

5 reasons for quality control

When does manual quality control make sense to validate the photos? How much time must be scheduled for this? How high is the acceptance rate as a rule?

Experience shows that the quality of images taken with Photo Collect is generally high and a large proportion of the images submitted can be used for photo ID. Responsible for this, in addition to the background removal and the automatic positioning of the face in the image, is a sophisticated quality assurance process which, in conjunction with automated quality checks and semi-automated quality control, enables high image quality even with high volumes.

Automatic quality checks

Automatic quality checks detect the most common errors as soon as the photos are uploaded and make it possible to immediately request a better photo. For example, pictures without people, group photos, people wearing sunglasses, faces taken too close, a sideways head position (looking left or right) or an unfavorable camera position (frog or bird's eye view) are automatically detected and objected to. The automatic quality checks are performed on all photos and lead to an automatic rejection of the first image in approx. 9.5% of the cases.

Manual quality control

After the automatic quality checks, manual quality control is another important element in the quality assurance of Photo Collect. Our data show that the rejection rate here can be reduced by three quarters through the automatic quality checks.

During manual quality control, each uploaded photo is briefly reviewed by a person and either accepted or rejected. If necessary, the image can also be edited directly in the workflow. Usually, this work is done by the customer themselves, often by the HR department, corporate security or the ID office. If a photo does not meet the requirements, it can be rejected with a predefined rejection reason (e.g., illumination or image quality). The person who uploaded the photo is then automatically informed of the rejection reason by e-mail or SMS and can submit a better photo later.

Quality control is optional and can be skipped if required. However, we highly recommend performing this step. Here are the five reasons why:

  1. Cost: It is much cheaper to reject a photo at this point than to have to re-produce the badge later because the photo is not satisfactory.
  2. Time: Rejected photos are usually quickly replaced with a new photo. Since communication is fully automated, this takes virtually no time. The quality control itself is very fast, we calculate less than one second per photo. Quality control can also be carried out in parallel in a team - this allows enormous quantities of photos to be checked in a very short time.
  3. Useful tips: With the rejection notice, the uploading person receives tips on how to avoid this very rejection. Was the illumination poor? Make sure there is sufficient light from the front and take the photo during the day. If the image quality is insufficient: clean the cell phone surface before taking the photo.
  4. High success rate: In 98% of the cases, the quality control results in a usable image. Usually 90% of all photos can be used on the first upload, 80% on the second attempt, and 50% on the third and final attempt. After that, the employee is usually captured via the Photo Collect app.
  5. Higher satisfaction: It's easier to submit the photo digitally on the second try than having to deal with capture in the card office or an analog submission. And with no media disruption, it's usually much faster, too.
  6. Manual post-processing: Okay, we only wanted to mention five reasons, but here's a bonus reason: In quality control, the images can be post-processed if necessary. Rotation, cropping, brightness and contrast can also still be adjusted manually - this is only done for 1% of all photos, but our customers appreciate the option of not having to use an external tool for this.

Conclusion

The quality checks and manual quality control are an efficient component in the Photo Collect process. We would be happy to show you how it works in detail and, if you like, also demonstrate the celebrity check, which can be used to easily and reliably identify and, if necessary, sort out images of celebrities sent in on a whim. Contact us for a demo!

More blog posts