by Jim Lundy
(Aragon Research) – At AWS re:Invent 2017, Amazon’s event last week in Las Vegas, it announced new capabilities that will make it more of a serious player in AI-based cloud computing.
Even more importantly, Amazon has clearly bet on content analytics and is pushing the envelope in video and voice analytics—two major tenants of the overall Intelligent Content Analytics (ICA) market, as defined by Aragon. (Note: Amazon announced many new cloud services tied to AWS, which we will cover separately.)
To clarify, Amazon Rekognition is not new—it is a deep learning service that is focused on image and video analytics. To do this, however, most of the processing needs to be done on images and videos stored in Amazon AWS.
The new Amazon DeepLens video camera.
The key news:
- Amazon Rekognition Video analytics can do real-time image recognition in video feeds, making it able to recognize people in a video. This is a game changer and we would note that video analytics is leaving the security and defense industries and is going mainstream.
- Amazon Transcribe provides auto-transcription of voice files, making it an important tool for indexing and then being able to analyze voice and video recordings (that contain audio).
- Amazon Comprehend is a sentiment analysis service. Sentiment is critical to improving customer journeys and has been something known in contact centers. Now, new services and applications that use Comprehend can react immediately to negative sentiment.
- Amazon Translate can do auto-translation of languages in real time. While in preview only now, Translate is expected to compete with Google—which has had these capabilities for years. We expect that people building chatbots with Amazon Lex will use Amazon Translate.
Amazon DeepLens Is Deep Learning and Video Combined
All of this combined with the new Amazon DeepLens video camera means that our concept of robotic telepresence that we defined in 2013 is coming to life.
DeepLens is focused on developers and it interacts with Rekognition algorithms. Amazon hopes that developers will choose its service, not the Google Clips camera to develop new applications using AWS Greengrass and AWS Lambda services.
The new Google Clips camera.
Clips vs. DeepLens
If there is any doubt that video analytics is going mainstream, one only has to look at Google and Amazon, both in a race to ship their cameras.
For the Amazon DeepLens, one of the things that is sure to please developers is that the camera is already pre-trained to do facial recognition.
The Google Clips offering, which is also still in preview mode, looks more ready for consumer use and it also has built-in facial recognition. We think that both of these prototype cameras represent the first generation. The moves from Google and Amazon will be the first of many efforts to leverage real-time video analytics.
Bottom Line
There is no doubt that video analytics will be a boon to the enterprise. The possibilities are limitless when we look at how video analytics and overall content analytics can be put to use.