Amazon Rekognition API
Image and video analysis using machine learning and artificial intelligence.
Facial analysis features, such as those available in Amazon Rekognition, allow users to identify faces in static images or videos, as well as determine their properties.
For example, Amazon Rekognition can analyze properties such as eyes (open or closed), mood, hair colour, and the visual geometry of a face.
The detected properties are extremely useful for clients who need to search among millions of images or order them within seconds by metadata tags (e.g. ‘funny’, ‘glasses’, ‘age range’) or to identify a specific person (i.e. when recognising faces from a source image or a unique identifier).
Clients: NFL, CBS, National Geographic, Marinus, SmugMug, SkyNews, Influential
- Content analysis using computer vision (CBS Corporation)
Solution: CBS places significant efforts to ensure they moderate inappropriate content within their programming as to not offend global viewers or violate government regulations. To scale internal processes, Amazon Rekognition is used to automate the moderation of video content while leveraging the new feature of Custom Labels to further refine moderation models.
Result: Amazon Rekognition enables CBS to automate the tagging of sensitive content such as nudity, obscene gestures, and violence, and speed up processing from hours to minutes.
- Improving the search effectiveness of user queries (Influential)
Solution: In addition to in-house AI/ML algorithms, Influential partners with third parties to enrich their datasets in order to better facilitate influencer sourcing. Amazon Rekognition object and scene detection allows Influential to better segment the influencer population into specific verticals and topics based on what media they post alongside their social media content.
Result: By extending search capabilities beyond just text, Influential allows for better training of their Brand Match Score, which when combined with Rekognition’s user-friendly tags & labels increase the hit-rate on user queries by over 200%
- PBS uses object detection to streamline media content operations
- Pinterest uses text detection to moderate user uploaded images
- Aella Credit uses facial recognition and comparison to verify the identity of customers.
- Thorn uses search based on Facial recognition to quickly identify minors who have been trafficked for sexual exploitation.
- The use of facial recognition and facial comparison of 75,000 students in the ICFES exam is a guarantee of a fair test.
- With Custom Labels, San Diego Gas & Electric identifies transformer damage using drone footage.
Simplify media analysis by automatically detecting empty frames, captions, changes of perspective and colour fades. By automating these tasks, you can reduce the time, effort and costs required to complete workflows such as video ad insertion, content analysis and content production.
Enable content retrieval
Amazon Rekognition automatically extracts metadata from image and video files that capture objects, faces, text and more. This metadata is used to search for images and videos based on keywords and help find the right means of syndicating content.
Tagging inappropriate content
Amazon Rekognition can automatically tag inappropriate content, such as images and videos that contain nudity, violence or weapons. Using the returned detailed metadata, you can create your own rules based on what’s considered acceptable by the user’s culture and demographic group.
Amazon Rekognition can be used to create scalable authentication workflows for automatic payments and other authentication scenarios. Amazon Rekognition provides an easy way to verify the faces of registered users by comparing a photo or selfie with an identity document (such as a driving licence).
Automate personal protective equipment (PPE) detection at all scales to improve workplace safety and ensure better health and safety compliance. With Amazon Rekognition, you can analyse camera images to determine if the people in the images are wearing PPE for their faces, hands and heads.
Identifying products, attractions and brands
App developers can use Amazon Rekognition Custom Labels to identify specific items in social media and photo apps. For example, a custom model can be trained to identify city landmarks from photos and provide tourists with information about their history, opening hours, and ticket prices.