Amazon Rekognition overview
Rekognition is an AWS SageMaker AI service for image recognition. Rekognition identifies objects, people, text, scenes and activities in images and video. Rekognition has both built in recognition capabilities and custom tags which allow you to label objects and people important to your business. When a tag, or label is identified Rekognition returns a percentage certainty so you can decide on the value of the inference.
Amazon Rekognition contains pre-trained algorithms that can be used immediately skipping the time consuming task of training image recognition algorithms. This pre-training is very useful when there little labeled data, or it would be expensive to develop a labeled corpus of images.
- AWS docs: https://aws.amazon.com/rekognition
- AWS FAQs: https://aws.amazon.com/rekognition/faqs
The Key features of Rekognition are:
- object and scene detection, image labling
- facial recognition
- facial analysis
- face comparison
- people paths (tracking people moving in a video)
- unsafe image detection
- celebrity recognition
- text in image
- personal protective equipment (PPE) detection
Video: What Is Amazon Rekognition?
This is a 1.42 minute from AWS. It is a brief overview of Rekognition features and use cases.
Custom image labling
Amazon Rekognition can be used to identify objects that are relevant to a specific knowledge domain business use case. Rekcognition is provided with a corpus of labeled images which is uses as training data for customised identification of objects.
Video: Rekognition Custom Labels – AWS Online Tech Talks
This is a 28.23 minute video from AWS.
The most common use-cases for Rekognition Image include:
- Searchable Image Library
- Face-Based User Verification
- Sentiment Analysis of images
- Facial Recognition
- Image Moderation
Image Moderation is also known as Content Moderation. It involves the detection of gore, drugs, explicit nudity, or suggestive nudity in an image to classify the content and enable it to be removed, or restricted to certain users, for example adults only.
The most common use-cases for Rekognition Video include:
- Search Index for video archives
- Easy filtering of video for explicit and suggestive content
- Tracking the activities of people
- Photo by Anne Nygård on Unsplash