Tech is cool and all, but what merit does it have if not solving the business problems?
The most effective solution to the business problem is where innovation overlaps with a proper business case. You can make the most complex algorithm, but if it doesn’t translate well to the business case — you won’t get far with that.
In this article, we will talk about the companies that rely on their expertise to help businesses solve their problems in the most efficient way.
AnyClip provides the intelligent tools that allow for managing video content libraries — and something beyond that.
Their solution analyzes the data extracted from video to enable businesses to effectively store, manage, and access the content from their libraries.
The videos get tagged with the help of the following tech stack:
- Optical character recognition;
- Neural networks;
- Digital image processing;
- Frame-by-frame object recognition.
The tagged content can then be easily searched through for things that were said or shown during the video.
Apart from that, AnyClip provides the tools that utilize the data from video content for targeting, monetization, automating recommendations, and tracking performance metrics.
So, AnyClip video analysis can serve a variety of businesses looking to optimize their operations: retail, e-commerce, publishing, and media.
But in terms of media-oriented solutions, AnyClip fails at providing the ability to act on the content it analyzes while nailing the distribution. You can’t use the metadata to edit the content for further use: for example, cut the highlights or crop the video to the portrait aspect ratio.
The way AnyClip approaches video content analysis is also far from perfect. Even though the object recognition technology provides some semantics around a video asset, it fails to deliver an end-to-end business automation solution.
The thing is the semantics is not the guy you come to with video. Object recognition technology works by decomposing the video into separate frames and treating the frames like images.
Share your thoughts on the article!
See where I’m going?
Such video analysis couldn’t understand the context of the video or its narrative structure even if its life depended on it. But I would argue that it is a crucial feature for a media-oriented solution.
In addition, the video analytics solution is actually going over the top with its methods: transforming the video into the semantics and back might be overkill for many tasks — those steps can be skipped for more effective production.
AIHunters delivers AI video analytics solutions that specifically aim at automating and optimizing the production process of the media & entertainment industry.
The solution provided by AIHunters doesn’t need preliminary training for each time different problems, being able to cover the widest range of use cases.
Cognitive computing algorithms implemented by AIHunters deliver brain-inspired computer vision capabilities that haven't existed before.
So, let’s say you need to analyze the footage from the sports type AIHunters hasn’t tried yet. It would take around 2 weeks to adapt the cognitive computing algorithm to do the task compared to the 6 months it would take the competitors to build the AI-driven solution from scratch.
Moreover, you do not just get probability scores — you get decisions that actually benefit your business.
The sophisticated tech allows CognitiveMill™ to augment the human editor in up to 70-100% of video-related production processes:
- Highlight, best-moments, and trailer generation;
- Cast management;
- End-credits detection;
- Cropping the footage to portrait aspect ratio;
- Nudity filtering
- Graphic overlay detection.
CognitiveMill is built in a way that allows it to be easily integrated into any video production pipeline without any additional setup. There is no UI or settings panel to tinker with — you just plug and play.
The system can also run multiple processes at the same time, so you won’t need to re-do the whole analysis for different purposes. For instance, you might want to:
- find a safe point to skip end-credits;
- recognize and tag actors in all the scenes;
- find the best moments and crop the for vertical aspect ratio,
Explore the media automation potential!
You can easily do that by running CognitiveMill just once.
Clarifai is more about a universal approach. The AI video analytics company provides an AI platform that can be adjusted to work with images, videos, text, and audio to extract important information.
The company heavily leans into building a customizable AI platform for a wide variety of purposes. It provides pre-build AI models that can be built-in in all kinds of software and further customized for data prep, insights discovery, pattern recognition.
Clarifai delivers its solutions to serve the whole spectrum of industries: from aviation and manufacturing to tourism and media.
With that many spheres of business to work with, I bet you can see the issue.
How do you act on the data you get from the AI models? You build the workflow by yourself.
Well, that’s not fun.
Valossa takes a step further with its video content analysis. Now you do not just get a tool that pulls the data from the video content, but the one that completely automates the production process.
The video analytics solution offers the ability to use the extracted data to automatically cut video content according to the set parameters. It can recognize the context of the video, so making a compilation of the tense moments in a reality show won’t be an issue.
The same goes for the promotional content. The tool can generate targeted videos catering to different demographics, increasing the effectiveness of the promotional campaign.
And the third major ability is automated content moderation. AI can go through videos and images to identify explicit content such as nudity, violence, substance use and filter them out.
Here, it’s the same old story: from the technological standpoint, the solution is quite straightforward. Here the tech is based on simple deep learning models along with object recognition algorithms.
So, you guessed it — Valossa also breaks the footage down into frames and analyzes each frame as if it was an image.
That won’t get you anywhere in terms of actually understanding the content. In turn, narrative intelligence is essential for correctly editing and tagging the content in accordance with certain purposes.
We optimize your workflows. You go on and be creative.
And again: the data remains the data. The data itself has no use for the media and entertainment industry. Probability scores and tags ain’t gonna do much if you need your cast members' information, a highlight compilation, or a trailer generation.
5. WSC Sports
WSC Sports focuses on sports-related video analysis solutions. They build technology that analyzes sporting events to provide broadcasters with more content to use.
How do they do it? Just like most of the competition — with their custom AI cloud platform. Plus, a generous amount of external data feed (events manually registered by scouts present on a match).
Using the cloud platform, sports broadcasters can generate personalized video content to cater to diverse audiences and increase their engagement with said content.
WSC Sports is focused on delivering automatically generated videos containing sports highlights. All of your dunks, goals, touchdowns, all of the drama and tense moments can be cut from the game footage and then broadcasted on TV or shared on social media.
And that focus is what is so limiting about the WSC approach to video analysis. The AI is trained well to recognize the best moments of the sporting event, but can’t do much else. For instance, if you want to create a trailer for a movie — you will have to re-train the AI model.
The way the footage is analyzed for highlights generation also leaves a lot to be desired. For the most part, WSC Sports relies on external data feeds to navigate the footage and look for the most exciting moments.
And here’s the best part: the highlight-worthy moments are found with the help of audio analysis. The AI looks for the cheering sounds and what not to tag the moments that go into the compilation.
But what about sports with less vocal audiences? We may never know.
In addition to that, the solution can not do more with the content. No safe skipping. no overlay recognition, no celebrity identification — there’s a lot that can be done to further automate the production.
Acting as a virtual assistant for the M&E industry, the tool thoroughly analyzes the footage and — you’ll be surprised — makes sense of it. The solution completely understands the narrative structure of the footage to pinpoint main and secondary characters, identify important moments, and so on.