How I met AIHunters

“We are AIHunters. We work with artificial intelligence, leverage AI and cognitive computing (СС) to enable cognitive business automation in the Media and Entertainment industry. We’ve created a unique cloud platform called Cognitive Mill™ that analyzes various video content...”

That’s what I was presented with when I decided to join AIHunters two months ago. EXCITING, isn’t it?

Well, it is until you have to sort out how it actually works.

I’ve never worked with this technology and knew nothing about the industry. But that was my chance to break new ground, the team was ready to teach me, and I was eager to learn from scratch.

And I think it was “the beginning of a beautiful friendship!”

AIHunters had already had an AI-powered, functioning platform with several ready-to-use products and almost no documentation. Everything about how it worked and improvement plans existed only in the creators' heads.

The team needed someone to introduce the daring ideas they made real to the world. It has become my mission.

This is my story about how complicated things, such as artificial intelligence and cognitive computing, can become simple in no time when you join a dream team as AIHunters is.

Step 1. A deep dive into cognitive computing and artificial intelligence

Before starting to learn about the product, I needed the context. I was binge-watching YouTube videos, binge-reading articles — anything that could help me understand what I was into.

At the same time, the team was telling me about the platform and, sure thing, about cognitive computing and artificial intelligence from the perspective of the platform.

I had to learn and I had to write — we needed the docs ASAP!

More and more information was poured into the ocean where again I was learning to swim, navigate, and choose the right directions. That would be mayhem without my team, they guided me at every step.

As there are no unified agreed-upon definitions for such terms as cognitive computing and artificial intelligence, first, I needed to figure out what meaning we at AIHunters put into these terms to make sure that we are all on the same page.

From what I’ve learned from various sources, both AI and CC often come side by side under the umbrella term of artificial intelligence. They definitely have much in common: they use similar technologies to achieve slightly different goals.

AI’s purpose is to create intelligent machines that can take action instead of humans.

CC aims to make intelligent machines or systems behave like people and make human-like decisions.

This is how I understood the concepts.

But it wasn’t exactly the way AIHunters interpreted these things. I had to look at AI and CC from a different angle.

And this is how it is from the AIHunter’s perspective. I’ll give the definitions first for you to see the bigger picture.

Artificial intelligence is about the technology that aims to create smart machines. AI-based technologies use algorithms, self-learning neural networks, computer vision, etc.

Cognitive computing is an add-on that complements artificial intelligence so that AI-powered machines are not only smart enough to solve specific problems, but they are able to make human-like decisions, complete tasks, and solve problems as humans do.

AI always comes together with two more popular terms that have become buzzwords: machine learning (ML) and deep learning (DL), while cognitive computing and AIHunters’ work are hard to imagine without three less trending but essential terms, such as:

• Cognitive science. A study of the human mind and intelligence of how a human mind functions and behaves.
• Probabilistic AI. A theoretical and practical approach that enables machines to analyze data and make predictions about new cases in unfamiliar circumstances based on the earlier acquired experience.
• Machine perception. The capability of the machine to perceive and interpret information like a human would through imitated human senses.

What is 100 percent true — artificial intelligence and cognitive computing perfectly complement each other and, when implemented together, can bring impressive results. One of such results is the creation of Cognitive Mill™.

Step 2. Cognitive Mill™: from idea to product

I wondered how it started and where we were heading.

It all began with an idea of how to implement automation in the Media & Entertainment industry that has so much content to process but requires too much time and human work.

The Team

The inspiring idea gathered the right people in the right place to make it real. Ideas turned into products. Products included various solutions for possible business cases.

I joined the team when it was all set. What impressed me the most was how engaged each team member was. For them, it wasn’t just a job, it was an essential part of life.

I’ve become a part of the team, and it has become a part of my life too, from now on, involved in something amazing.

Cognitive Mill™

Using a unique combination of the best AI practices with cognitive computing, the team created a system that is much more independent from the pre-taught scenarios. It makes informed decisions in any circumstances. And I was really impressed by how it works!

The AI concepts were the basis of the whole project that further developed into Cognitive Mill™ the way it is today.

So, what is Cognitive Mill™?

It is an AI-powered cloud computing platform reinforced with cognitive computing, the most innovative technologies, and complex math algorithms. It is able to process and analyze any type of video content and make human-like decisions depending on various business cases.

Can you imagine a cloud robot that imitates human cognitive abilities, ah?

It watches video content, analyzes the context, understands and summarizes it, and makes informed decisions instead of me or you. This is so amazing as if you witness a working miracle. You cannot see it, but you use the results of its work.

Video is complex to process. Machines equipped with AI are smart enough for object recognition or speech-to-text transcription, but cannot be trusted in their understanding of what they ‘hear’ and ‘see’. So, the decisions are still mostly taken by people. It’s a tiring, time-consuming, and expensive post-production routine. Cognitive Mill™ can ease it up.

Our virtual robot processes a video in two stages:

1. Representations stage. Where it watches a video and perceives the information like a human.

2. Cognitive decisions stage. Where it makes human-like decisions based on the analysis of the watched media.

Our AI scientists keep constantly improving the logic of the robot at both stages, which is actually the most challenging part.

You can interact with the platform via the API or the UI.

I really like the UI, it is concise and user-friendly. You can even change the color theme according to your mood!

Here you can create processes, check process status, download JSON files of the completed processes, check process results in the visualizer before compiling a trailer, for example.

There are two types of processes that I classify by the result you get after running them:

1. Metadata processes (like Soccer meta, Crop meta, etc.) that generate a JSON file with all the required metadata for further processing with third-party systems and software or with Cognitive Mill™ media generating product called Cognitive Mill™.

2. Media generation processes (like Media mill, Soccer trailer compilation) that provide a trailer.json file for further trailer generation with third-party systems and software and a trailer or a set of trailers that can be downloaded immediately.

Running a process is very easy, though the platform is complex. For most pipelines, all you need is a link to the video and a name you want to give to your process :)

In case you want to get acquainted with our pipeline workflow in more detail, kindly see Cognitive Computing For The Media & Entertainment Industry Automation.

After the process is completed as the result of the generated decision, we get a JSON file containing the metadata necessary for post-production. It always contains segments with time markers of the specific events, for example, highlights, side content that can be skipped, cropping data, etc.

Step 3. Cognitive Mill™ products and solutions

When I joined AIHunters, Cognitive Mill™ already had six products and even more amazing solutions. I was surprised at how cool it was and I can’t wait to share it with you.

From the product’s name you can easily guess what its purpose is, like, for example, Cognitive Mill™ for skipping the unessential content or Cognitive Mill™ for nude detection.

You will find exhaustive information about all the products on our website cognitivemill.com.

And I want to tell you about the ones that impressed me the most.

Cognitive Mill™

Cognitive Mill™ is actually two products in one, one was designed for sports highlights and the other one for movies:

• Cognitive Mill™ for sports such as soccer, hockey, basketball, American football, and boxing. But the ready-made pipelines can be easily adapted to any other kind of sport.
• Cognitive Mill™ for automated movie trailer generation. This is an amazing tool for creating your own trailers for any movie or series episode in the fully automated mode.
  Here’s an example of one of the episodes of a Korean series that went viral in 2021, Squid Game.

Cognitive Mill™

Cognitive Mill™ adapts horizontal videos to the portrait aspect ratio. It’s very handy, right? I bet you also watch most of the content on your mobile phones and share stuff in TikTok and Instagram Stories. That’s why the default aspect ratio of the cropping frame is 9:16, by the way.

Just look how the cropping frame focuses on Billie Eilish even in the dark! The team says that the pipeline has been created with the help of machine perception and imitation of human focus without DL, which makes it universal.

Cognitive Mill™

Cognitive Mill™ is strong at video summarization. The system identifies main and secondary characters. It leaves no doubt who is the main character and who is the secondary character and provides information about the ones you select.

You can even generate a trailer that includes only those scenes where the selected character appears.

You can see how it works and try it out in our Demo account at run.cognitivemill.com and create your own account and run processes yourselves with trial quotas.

Wrapping up: achievements and threats

What we’ve done

Cognitive Mill™ is pretty young, only one year and three months old. And it has already achieved impressive results. For example, the platform processes media 50 times faster than a human. Sounds good, doesn’t it?

We already process more than 100 hours of media per day, but we aim to increase it to 2000 hours of media daily.

And just imagine a team of fewer than 20 people that within a year, even less, created more than 50 unique AI algorithms! I’ve never seen them, but believe me, our AI scientists can be trusted. 16 cognitive automation pipelines have been created. At least 8 of them are working, I can guarantee, I’ve tried them out!

Is it dangerous?

Some fear the rise of the machines and the future pictured in sci-fi movies like The Terminator or Ex Machina. Others fear that machines will replace people in many jobs.
But, on the other hand, technology drives development. Many people keep constantly learning to keep up with modern life, which is definitely good for them. I’m among those constantly learning individuals too, and I like it!

We at AIHunters look on the bright side.

We consider AI and CC aids to assist people where the volume is huge while time and knowledge are limited and only then replace them when people themselves don’t want to waste time on monotonous work deprived of creativity.

Cognitive Mill™ is in no way a threat — it is an assistant, and its purpose is to help. It was created to make life easier and save time for something beautiful.

We’ve created Cognitive Mill™ to make this world a better and more understandable place.