It has been a really busy period for us at Plastream. We were working closely with one of the biggest TV broadcasters on an in-video analysis project for their digital TV player online. We were in charge of mass-tagging episodes of popular series and matching detected items with similar or identical e-commerce products.
Let us get you through the overall and show how Plastream can be implemented for improving some processes.
What is our project about?
In collaboration with a publisher, we analysed a few episodes of well-known and popular TV series. Based on received materials, we prepared an in-depth analysis of items shown in each episode and then, thanks to our technology, we automatically matched it with products found in e-commerce stores connected to the platform, giving the end user possibility of purchasing the product right here and right now in a digital TV player in the Internet.
How did the process look like?
- We received video materials from the publisher and uploaded them to our platform for in-depth analysis.
- Then, we divided each episode into very short sections, even less than second-long ones.
- We grouped those tracks which included the same clothing item – it is common in some episodes that one person appears in the same clothing set for the whole duration of the episode.
- Tracks were matched with similar or identical products in e-commerce stores. It’s worth mentioning it has never been just one item: Plastream made it find much more clothes matching the item on the track.
- The material was released on publisher’s channels: online TV player – wherever the user paused the video, they were served a contextual advert with clothes matching the item in that particular moment. This rule applied to the whole video.
- We were analysing and optimizing the performance when needed.
We deliver technology that speeds up the process that publishers need to make anyway if they consider monetizing their assets.
What is in-video analytics?
According to Cisco, by 2021 video will be 82% of all consumer Internet traffic. Online video is considered the fastest growing segment within the digital advertising landscape, and both advertisers and publishers find it appealing.
However, there is so much video content created nowadays that publishers face a challenge of keeping up with monetizing it. One of the most popular ways of monetizing for publishers is to link content in-video with e-commerce. Up to the date, publishers rather do it manually and slower than technology can already do.
In-video analytics is about detecting items in video content and matching it with products in e-commerce stores. It can result in higher retail sales conversion, better user experience and higher customer satisfaction – and all sides of the transaction can benefit from it.
In-video commerce involves users in a purchase process even if the purchase is not finalized. The user gets familiar with an advertisement that is paired with the context in the video, so the message hits stronger and with more awereness than a traditional advert. Digital TV achieves their commercial goals thanks to a better offer for advertisers.
How does Plastream work?
Plastream uses neural networks, artificial intelligence and machine learning. Based on the delivered video, we give accurate recommendations (using Similarity Search) of similar or identical products in e-commerce stores.
On the other hand, Plastream can classify and detect products placed in own productions made by a particular TV broadcaster.
The mechanism analyses the content, detects items and based on searched elements it is able to match the video content with products available online or with a selected advertised product. The analysis is very precise – it detects not only the shape, design or colour but also texture, material and characteristics.
Benefits of collaborating with Plastream
Using context for monetization
Context ads are flipping ad systems that still had a chance of being just a few years ago. Today, in order to get a recipient’s attention you have to strive for it in an industrial manner. All commercial context must go in pair with the content provided by the publisher. Example- next to pictures of celebrities in a yellow dress should be recommendations of red dresses available in e-commerce stores. Context must be as “slim” as possible in order to give expected results and moved on to a monetized aspect.
Plastream offers technology that analyzes video material with the precision that recognizes individual products. We believe that matching context gives better results – CTR on the level of 10-20% with extremely accurate recommendations.
Helping in archiving and creating new content
A similar issue concerns archiving that takes place manually, it’s very time-consuming and not always precise. Plastream is able to tag video material based on its context which makes archiving easier and allows a faster material search that can be used to produce new materials.
Don’t hesitate to contact us!
Advertisers who are interested in publishing and recommending their assortment on our platform can do so by contacting us beforehand.