AWS Resources

Goal

Create the AWS resources the backend will rely on:

S3 for raw uploads and thumbnails
DynamoDB for photos, people, and occurrences
Rekognition Collection for face matching

Use one region everywhere

Create ALL resources in the same region (recommended: us-east-1). Rekognition collections are region-specific, and mismatches are a common failure.

Minimum prerequisites

Make sure you have:

AWS account access
AWS CLI configured and working
A single AWS region chosen (use the same region for everything in v2)

Quick check (CLI):

aws sts get-caller-identity should return your account details.

AWS CLI setup

This project uses AWS CLI for Rekognition collection creation and verification. If your AWS CLI is not configured, complete this section before continuing.

Step 1: Install AWS CLI

Download and install AWS CLI v2 for your OS from the official AWS site.
After installation, verify:

aws --version

You should see output like: aws-cli/2.x.x ...

Step 2: Configure credentials

Run:

aws configure

Enter:

AWS Access Key ID: (from your IAM user)
AWS Secret Access Key: (from your IAM user)
Default region name: us-east-1
Default output format: json

Recommended default region

Set the default region to us-east-1 so you don’t forget the region in commands.

Step 3: Verify you're authenticated

Run:

aws sts get-caller-identity

Expected output contains your:

Account
Arn
UserId

If this fails, your credentials are not set up correctly.

Service Map

S3

We will use two buckets:

Raw bucket (prefix: photos-raw/) : stores original uploads
Thumbnails bucket (prefix: faces-thumbs/): stores cropped face thumbnails and processed outputs

DynamoDB

We will create three tables:

Photos table: one record per uploaded photo
Persons table: one record per person cluster
Occurrences table: mapping of person ↔ photo with face metadata

Rekognition

We will create one Collection:

beetroot-faces

Note: S3 “folders” are just prefixes. We'll create them by uploading a placeholder object if needed.

Create resources (step-by-step)

Create bucket : beetroot-raw
1. Go to S3 → Buckets → Create bucket
2. Bucket name: beetroot-raw
3. AWS Region: us-east-1
4. Block all public access: ON
5. Versioning: ON
6. Default encryption: ON (SSE-S3 is fine)
7. Click Create bucket
Create bucket : beetroot-thumbs
1. Go to S3 → Buckets → Create bucket
2. Bucket name: beetroot-thumbs
3. AWS Region: us-east-1
4. Block all public access: ON
5. Versioning: ON
6. Default encryption: ON (SSE-S3 is fine)
7. Click Create bucket

Naming tip

S3 bucket names must be globally unique.

Create table : Persons
- Go to DynamoDB → Tables → Create table
- Table name: Persons
- Partition key: personId (String)
- Capacity mode: On-demand
- Click Create table
Create table : Photos
- Go to DynamoDB → Tables → Create table
- Table name: Photos
- Partition key: photoId (String)
- Capacity mode: On-demand
- Click Create table
Create table : Occurrences
- Go to DynamoDB → Tables → Create table
- Table name: Occurrences
- Partition key: personId (String)
- Sort key: photoId (String)
- Capacity mode: On-demand
- Click Create table

Why this key design?

This design makes the later API query simple:

List people → read from Persons
Photos for a person → query Occurrences by personId

Why CLI here?

Rekognition collections cannot be created via the AWS Console, they're only available through the CLI or SDK.

Skip step 1 if you have AWS CLI installed in your device.

Open CloudShell in us-east-1
- In the AWS Console top bar, open CloudShell
- Ensure the region is us-east-1

Create the collection

Run:

aws rekognition create-collection `
  --collection-id "beetroot-faces" `
  --region us-east-1

Verify the collection exists
Run:
```
aws rekognition list-collections --region us-east-1
```
Confirm you see: beetroot-faces

Checkpoint

By now, you should have:

2 S3 buckets (raw + thumbs)
3 DynamoDB tables (Photos, People, Occurrences)
1 Rekognition collection

AWS Resources

On this page