One of the magical features of our free photo platform is the automated tagging of people based on image processing of bib numbers. For each photo that is uploaded, we automatically send it to a scanning service and it returns the bib numbers. We add nice little touches like filtering for bibs less than 1000 so you can remove incomplete tags.
As part of our Photo 2.0 upgrade, we have moved from Google Vision doing the tagging to Tagily, which is a more optimized bib tag recognition system. They use AI and machine learning to be able to better recognize and read bib numbers. Here is an example of a photo with bib numbers that were automatically found:
The tagging happens in a matter of seconds of upload of the photos. This means participants and spectators can find photos of themselves and the others they follow quickly. Coupled with watermarking of photos for sponsors, this can really increase the engagement factor for your event and for your sponsors.
If you have marked your event as a cycling event, then we will automatically use Tagily’s cycling bib number mode. The benefit of their machine learning capability is they have used millions of photos to train their algorithm. This is going to be very useful as more and more cycling and gravel events are using our platform.
Other Vision Technology
We also use Amazon’s Rekognition on each photo uploaded to our platform to scan for inappropriate images. These get put into the manual review queue for the event director to approve. This is important as we have enabled events to have the ability to allow participants to upload photos.
Also, note the limitations of the vision technology: – no letters (except F in front, for Feminine), no symbols (except a dash before the last digit), and a max 5 digits.