AWS Machine Learning Blog

VidMob combines computer vision and language AI services for data-driven creative asset production

VidMob is a social video creation platform that marketers of all sizes can use to develop personalized advertising communications at scale. VidMob uses machine learning (ML) to power its SaaS application. This application uses metadata extraction and sentiment analysis to provide marketers with actionable insights into which creative assets resonate with their intended audience, and more importantly, why they resonate. For example, you can see how the placement and timing of text in a video ad influenced viewer engagement by reviewing data, such as the drop-off rate.

VidMob identified a challenge that marketers have when they drive brand and sales with their digital advertising efforts: They are often forced to make creative decisions without proper creative insights. With no actionable best practices, video marketers were often in the dark as to why certain campaigns worked while others didn’t. This drove VidMob to create the Agile Creative Suite (ACS), which leverages Amazon Rekognition Video, a deep-learning-based video analysis service. The ACS uses Amazon Rekognition Video for automatic metadata extraction and sentiment analysis, which provides you with the insights you need to understand how visual aspects of a video ad affect critical metrics, such as viewer retention.

These insights are seamlessly connected with a production platform that enables you to immediately incorporate insights when you revise your creative assets. By fully connecting the loop of production, data-driven insights, optimization, and re-publishing, the ACS is able to dramatically improve a brand’s overall return on marketing investment. The ACS also uses Amazon Rekognition Video to enable clients to organize and search their assets by key visual or language attributes.

As VidMob was developing the ACS, they discovered they would need a deep learning and computer vision service to power key features. Their research led them to many different services, but ultimately they determined that the speed and accuracy offered by Amazon Rekognition Video was unparalleled when compared to competitors.

“Amazon Rekognition has unlocked for us the tools to provide more precise and actionable creative learnings quickly and efficiently,” said Alex Collmer, CEO of VidMob. “AWS’s deep learning capabilities help to power our Agile Creative Suite, which finally provides marketers the insights they need to understand and improve their creative assets. We look forward to our continued partnership with AWS to continue to develop the most innovative tools to power creativity.”

How it works: VidMob and Amazon Rekognition Video

As an example, let’s look at an eight-second VidMob video ad. This video has been uploaded to the ACS, which displays the video metrics and allows users to filter their view by emotions, objects, people, and words. This capability is powered by machine learning on AWS and enables marketers to dive deep into the factors of the video that drive viewer retention.

The video uploaded to ACS is analyzed by Amazon Rekognition Video and then displays what people, objects, and emotions were detected. This short video example shows that Rekognition Video captured smile and surprised for emotions; and head, poster, person, envelope, mime, and performer for objects. Although this is a very short video, there are valuable insights that customers can use to formulate best practices and apply to subsequent video ads. For example, a customer using the ACS could filter for emotions in this particular video ad and notice that the drop-off rate coincides with the length of the surprised emotion.

VidMob also utilizes Amazon Comprehend and Amazon Transcribe to power their Agile Creative Suite. This combination enables VidMob to build high-quality machine learning language analysis into the ACS. According to Collmer, “We are able to transcribe text from video content, and quickly analyze it using Amazon Comprehend, allowing us to surface actionable insights to both our creator community and our clients, giving them a strategic edge in the market.”

The machine learning services on AWS that power VidMob’s Agile Creative Suite provide marketers and advertisers the data they’ve needed to start making creative decisions that are data-driven. “Video creators can now understand, for example, how the timing and sizing of text or logos influenced the performance of their creative assets,” says Collmer. “These types of insights illustrate why we truly believe that the role of AI is to empower and enhance human creativity. We are excited about our relationship with AWS, and as we continue integrating machine learning into our creative solutions, we have every expectation that AWS will be our machine learning platform of choice.”

To get started with Amazon Rekognition Video, see these posts:

Automated video editing with you as the star!

To get started with Amazon Comprehend, see these posts:

ClearView Social uses Amazon Comprehend to measure the impact of social sharing

Build a social media dashboard using Amazon Comprehend and BI services


About the Author

Kaiser Larsen is a Product Marketing Manager for AWS artificial intelligence solutions. Outside of work you’ll find him hiking, cooking for family and friends, and eating ice cream whenever there’s an excuse to celebrate.