Considerations - Auto Check-In App



This guide focuses solely on architectural and technical issues related to developing and deploying the Auto Check-In App. It does not cover non-architectural or technical considerations that may be associated with this use case, such as legal and privacy issues. For example, in certain geographies there may be laws regarding the collection, storage, and processing of biometric information. You should review legal requirements in your jurisdiction and obtain legal advice where appropriate.

In addition, there are a number of privacy and transparency considerations around the use of facial comparison in commercial settings. You should educate yourself on these issues and take them into account when designing and deploying your application. We recommend that you provide information to end users about how the application works and how data collected from the application will be used, stored, and retained, and offer an alternate verification option for end users who do not wish to use the application. For more information, review the following additional resources and best practices:

Amazon Rekognition

Amazon Rekognition uses deep learning models to perform face detection and to search for faces in collections. It continues to improve the accuracy of its models based on customer feedback and advances in deep learning research. These improvements are shipped as model updates.

When you launch the Auto Check-In App solution, the face collection that is created is associated with the most recent version of the model. If a new version of the model is released, this solution’s collection will continue to use the model that is already associated with the collection. For more information, see Model Versioning in the Amazon Rekognition Developer Guide.

Similarity Threshold

This solution uses a similarity threshold to determine whether two faces have high similarity. The threshold controls how many results are returned based on the similarity to the face being matched. Responses that are below the threshold are not returned.

The threshold is a value between 0 and 100. The default threshold is 99. For more information, see Searching for Faces Within a Collection in the Amazon Rekognition Developer Guide.

Registration Image Preparation

During event registration, allow event attendees who want to use facial comparison during check-in to send an image of their face. After an attendee sends the image, make sure that the image is formatted correctly. The image should be in .jpg, .jpeg, or .png format, and the file name should be the name that you expect to be displayed during the event. For example, jeff.jpg or andy.png. Also, make sure that the attendee's face is the only face in the image. For additional information, see Recommendation for Facial Recognition Input Images in the Amazon Rekognition Developer Guide. Then, upload the photo to the solution’s Amazon Simple Storage Service (Amazon S3) bucket.

Event Image Preparation

To help ensure the most accurate results, make sure that your attendees’ faces are sufficiently lit, and that your images are larger than 80x80 pixels. We recommend using an additional light to ensure the face is bright enough. For more information, see Limits in Amazon Rekognition in the Amazon Rekognition Developer Guide.

To reduce the latency from Amazon Rekognition and ensure the correct face is indexed, we also recommend cropping the facial images you capture at the event before sending them to Amazon API Gateway. By cropping the image, you can reduce the response time and remove unintended faces from the image. For more information, see Amazon Rekognition Image Operation Latency in the Amazon Rekognition Developer Guide.

To help prevent the solution from indexing unintended faces, this solution indexes the largest face in the image. All other faces in the image are not indexed.

Image Storage

After Amazon Rekognition extracts the facial features from a registration photo into a feature vector and adds the vector to the face collection, the registration photo is automatically deleted from Amazon S3. The photo taken at the event is not stored. The solution does not store any facial images after they are processed.

Processing Status

This solution includes the following status.

  • Stop here: No face is detected

  • Checking: A face is detected and being compared to faces in the face collection

  • Welcome: A face is found with high similarity

While a facial image is being processed, the status will change from Stop here to Checking. If a face with similarity above the threshold you defined during initial deployment, the status will change to Welcome. If no faces with similarity above the threshold are found, the status will stay in the Checking state.

Event Security

For customers who use this solution for events with additional security requirements, we recommend using multi-factor authentication during check-in as an additional layer of security. For example, in addition to the face matching, ask attendees to show their government-issued ID.

This solution does not provide anti-spoofing or liveness detection. You can, however, customize the solution to add this functionality. For example, you can send attendees a temporary code to use in addition to the picture at the event. You can also build a challenge-based workflow where the attendee has to perform a series of randomized steps such as smiling, tilting their head, or closing their eyes to verify liveness.

Operator Management

The Auto Check-In App includes a Shell script to grant you access to the solution’s UI. The UI does not provide operator administration. To add additional operators, you can use the included script. For more information, see Step 3.


We recommend testing the functionality of this solution before your event. We also recommend using the laptop and camera you plan to bring to the event.

Test Configuration

While you can use any laptop, camera, and monitor with the Auto Check-In App, we have tested this solution using an Apple MacBook Pro, a LogiTech C920S webcam, and a Dell P2419H monitor. We recommend changing the monitor settings to portrait mode and putting the camera on top of the monitor. We also recommend using an additional light to ensure the faces are well lit.