Amazon Rekognition charges you each time you analyze an image and for each set of facial feature vectors you store. You can store these facial feature vectors, known as face metadata, in collections to perform face searches.
Free Tier
As part of AWS’s Free Tier, you can get started with Amazon Rekognition for free. Upon sign-up, new Amazon Rekognition customers can analyze 5,000 images per month and store up to 1,000 face metadata each month, for the first 12 months.
Amazon Rekognition is now generally available
Try Amazon Rekognition for FreeOr Sign In to the Console
Upon sign-up, new Amazon Rekognition customers can analyze 5,000 images per month and store up to 1,000 face metadata each month, for the first 12 months.
Let’s assume your application analyzes 1 million images per month that require label detection.
You would use Amazon Rekognition’s DetectLabels APIs to analyze these 1 million images.
The price for 0 – 1 million images processed is $1.00 for 1,000 images (or $1.00/1,000 = $0.001 per image).
Total charge for each month = 1,000,000 * $0.001 = $1,000.00
Let’s assume your application analyzes 1 million images per month for both labels and faces. You would use the following number of images processed:
1 million DetectLabels API calls for label detection, and
1 million DetectFaces API calls for face detection.
Total images processed = 2 million.
The price for 0 – 1 million images processed per month is $1.00 per 1,000 images processed (or $1.00/1,000 = $0.001 per image processed).
The price for 1 – 10 million images processed per month is $0.80 per 1,000 images processed (or $0.80/1,000 = $0.0008 per image processed).
Total charge for each month = ($0.001*1,000,000) + ($0.0008*1,000,000) = $1,000.00 + $800.00 = $1,800.00
Let’s assume your application analyzes 40 million images to detect labels, analyze faces, and create a collection of 10 million faces. Thereafter, in the same month, you perform 1 million searches in the face collection you created, and delete it after 15 days.
You would use the following number of images processed:
40 million DetectLabels API calls for label detection,
40 million DetectFaces API calls for face detection,
40 million IndexFaces API calls to create a collection for face storage, and
1 million SearchFaces API calls for performing searches.
Total number of images processed = 40M + 40M + 40M + 1M = 121M
The price for images processed per month:
0 – 1 million images processed per month is $1.00 per 1,000 images processed (or $1.00/1,000 = $0.001 per image processed).
1 – 10 million images processed per month is $0.80 per 1,000 images processed (or $0.80/1,000 = $0.0008 per image processed).
10 – 100 million images processed per month is $0.60 per 1,000 images processed (or $0.60/1,000 = $0.0006 per image processed).
Over 100 million images processed per month is $0.80 per 1,000 images processed (or $0.40/1,000 = $0.0004 per image processed).
Charge for images processed = ($0.001*1,000,000) + ($0.0008*9,000,000) + ($0.0006*90,000,000) + ($0.0004*21,000,000) = $1,000.00 + $7,200.00 + $54,000.00 + $8,400.00 = $70,600.00
The price for face storage is $0.01 per 1,000 faces stored per month (or $0.01/1,000 = $0.00001 per face). Face storage is billed based on the number of days in the month. Assuming a 30-day month, the price for storing 1 face for 15 days is $0.000005 = ($0.00001/30)*15.
Charge for storing 10 million faces for 15 days = $0.000005*10,000,000 = $50.00
Total charge = Charge for images processed + Charge for storage = $70,600.00 + $50.00 = $70,650.00
