
AWS Rekognition: Turning Attendance Into Care
How we explored Amazon Rekognition to build a privacy-conscious attendance system focused on care, connection, and responsible AI.
Sometimes, a good application of technology doesn’t start with a massive business problem.
It starts with a simple question:
How can we take better care of the people around us?
Recently, we had the opportunity to test a different idea in a Catholic community here in São Carlos, Brazil: using AWS to better understand community participation and, most importantly, help people stay connected.
The idea was to use Amazon Rekognition as part of a voluntary attendance identification system.
How the idea works
At a high level, the flow is relatively simple.
Images are processed by Amazon Rekognition, which can detect faces and compare facial characteristics against previously registered references.
Based on the result and the confidence score returned by the service, our application can register attendance without relying on manual lists or QR codes.
The basic flow looks like this:
- A participant voluntarily registers a facial reference.
- The reference is added to a Rekognition collection.
- Images captured by the application are analyzed.
- Rekognition searches for a possible match.
- Our application evaluates the confidence score.
- If the result meets the defined criteria, attendance can be registered.
From an architecture perspective, the concept is simple.
The interesting part starts when you move from a diagram to the real world.
What we learned using Amazon Rekognition
During our tests, we were able to explore the complete basic workflow in practice:
- Creating and managing face collections
- Registering facial references
- Processing images
- Searching for faces inside a collection
- Working with confidence scores
- Handling the results returned by AWS
One of our biggest learnings was that facial recognition is not simply a binary question of:
"Is this the same person or not?"
Real environments introduce several variables.
Lighting changes.
People look at the camera from different angles.
Image quality changes.
A person may be farther away from the camera.
And even when Rekognition returns a potential match, the application still needs to decide what level of confidence is acceptable.
This made us think carefully about false positives, confidence thresholds, and situations where the system should request additional confirmation instead of automatically registering attendance.
What started as a relatively simple experiment quickly became an interesting exercise in cloud architecture, computer vision, Machine Learning, and responsible technology.
Attendance as a signal for care
But the goal was never to monitor who attended church or who did not.
The idea is to transform attendance into a possible signal for care.
Imagine someone who normally participates every week.
If that person suddenly stops attending for some time, this information could help someone from the community notice and reach out:
"Is everything okay?"
"Do you need anything?"
"Can we help?"
Not to enforce attendance.
Not to create rankings.
Not to judge participation.
But simply to notice people.
For me, that changes the entire perspective of the project.
The technology is not the final objective.
The human connection is.
Privacy must be part of the architecture
Another important lesson from this experiment is that when facial recognition is involved, privacy cannot be an afterthought.
It needs to be part of the solution from the beginning.
Some principles we considered essential include:
- Voluntary participation
- Clear communication about how the system works
- Explicit consent
- Restricted access to information
- Storing only what is truly necessary
- Defined retention policies
- Avoiding unnecessary exposure of biometric information
A technically impressive system that ignores privacy is not a successful system.
Responsible architecture means thinking about both what technology can do and what it should do.
Cloud technology can also solve very human problems
We usually associate cloud computing with scale.
Millions of requests.
Distributed systems.
Automation.
Performance.
Large organizations.
But one of the things I enjoy most about working with AWS is realizing that the same technologies can also be applied to extremely local problems.
Sometimes innovation does not mean building something bigger.
Sometimes it means using technology to make a community a little closer.
Maybe that is another definition of innovation:
Taking something extremely technical and using it to create something extremely human.
AWS does not always need to sit behind applications serving millions of people.
Sometimes, it can simply help a community notice when one person is missing.

Image disclosure: The images used in this article were generated with AI exclusively to illustrate what the system and its real-world experience could look like. No real images, faces, identities, attendance records, or biometric data from members of the community are exposed in this article.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article