Human machine
collaboration

AIHunters extends human cognitive abilities and makes media consumption and viewing experience as personalised as it can reach.

Cloud Augmented Intelligence
Cloud Augmented Intelligence Video analysis for Media & Entertainment Industry
Open-source software
Open-source software Our contribution to the OSS community
See an example of how AI works

Open-source
software products

We are keen to make the world a better, more understandable place. That’s why we share our innovation on all the levels from developer to enterprise.

AcurusTrack
AcurusTrack

A multi-object tracking component. Works in the conditions where identification and classical object trackers don’t (e.g. shaky/unstable camera footage, occlusions, motion blur, covered faces, etc.). Works on any object despite their nature.

 EvenVizion
EvenVizion

Camera localization and video stabilization component. Fixed coordinates system. Using Python and OpenCV.

SAAS. Augmented
intelligence in cloud

Our systems are tailor-made for the media and entertainment industry, knowing where the bottlenecks are and how to eliminate them with the power of human-AI partnership.

Movie credits detection

AI-based service for defining the beginning of the final credits in movies and TV series.

Football slow-motion replays detection

AI-based graphic elements detection in sports.

Emotional Journey analysis

Automated analysis of the emotional part of the movie.

Plagiarism detection and anti-piracy analysis

AI-based scene comparison to find contextual and visual similarities in video content.

Content compliance proof

Automatically detecting potentially harmful content by a single example.

Automated sports highlights generation

AI-based attention moments segmentation for professional football and basketball games.

Ads detection

Unsupervised detection of outstanding blocks in linear video streams.

Movie metadata generation

Automated movie context description — making metadata granulation hundreds of times deeper.

Video clustering

Automated video segmentation into logically-split parts

Unsupervised abnormal content (behaviour) analysis

Automatically detect pieces of content that differ from the pattern that the algorithm learned before.

Drag to Navigate

Contacts

We are AI Scientists, researchers, enablers. We strive to make the world a better, more understandable place. There’s enough entropy in the world. We want to reduce the amount.

Customer services
Support service

Our offices

  • Žalgirio av. 90 Fl.6D
  • Vilnius
  • 09303
  • Lithuania
  • Prospekt Dzerzhinskogo 57
  • Minsk
  • 220089
  • Belarus

Thank you for sharing your thoughts.
We appreciate your feedback!

We’d love to hear from you Send us a message and we’ll respond
as soon as possible
Movie credits detection

Find closing credits automatically

We use the pattern analysis AI and a number of custom configurable parameters to keep users engaged even when a movie is about to end.

Avoid false positives

The logic we created helps us distinguish between movie credits and other pieces of text, leaving the latter out. It works even on such specific cases as movie credits interrupted by another scene or post-credits scenes.

Offer more — fix your user retention

Suggest watching another movie of the same kind when end credits appear on screen. Make sure your users don’t leave by keeping them engaged.

Learn more
Automated sports highlights generation

Frame-precise scene segmentation

We apply scene-level contextual analysis to find the most thrilling and decisive moments of the game. Unsupervised learning ensures frame-precise scene segmentation and creation of a perfect highlight reel.

Adjustable to the specific needs

Whatever your business goal is, we can finetune highlights creation to it. You can simply tell the system what length, dynamics, angles, players, etc. do you need to include — and it will create a perfect highlight for you.

Ready in a matter of seconds

The system analyzes the dynamics of the game and decides on how to convey the intensity and angles of the game to make a highlight the most exciting. Highlights compilation will be ready seconds after the game is played, bringing you to the forefront of sports broadcasting.

Learn more
Football slow-motion replays detection

Detect branded graphic add-ins

Our lightweight service finds branded graphics of popular football leagues to capture the best moments (replays and other output of post-production) enclosed by these graphic elements.

On-the-spot detection

Find graphics in real-time by taking advantage of the AI-based monitoring of a stream. Analyze professional football matches and instantly find the best moments in them.

Boost sports production

Hyper-charge your sports production by leveraging the best of work done by the field production team.

Ads detection

Find ads in a stream

The system analyzes video content on a scene level and understands it as a human would do. Ads are detected automatically giving you full control over what the viewers are watching.

EPG correction

Synchronize what you are broadcasting with the schedule for current and upcoming programs. AI-based monitoring of a stream allows finding all the schedule inconsistencies to fix them.

Detect anything outstanding

AI-based monitoring of a stream allows finding all the interruptions and blocks that get out of context. After a frame-perfect detection of a particular block you decide what to show and what to leave behind.

Emotional Journey analysis

Detect emotions automatically

Leverage automated analysis of the emotional segment (in opposition to frame by frame imprecise analysis). Know what emotions appear on the persons’ faces and how are those emotions similar to other videos.

Decode the feel of the content

Dig deeper into video content by applying yet another level of decoding of what’s within. You can rely not only on textual descriptions, but also decode the feeling a video gives.

Recommend by vibe

Power your recommendation engine up by suggesting video content with the same characteristics emotions-wise. Give your viewers exactly what is needed at the moment.

Movie metadata generation

The most innovative technologies

Self-supervised, metric learning and representation learning approaches for manifold learning methods and unsupervised data recognition ensure deepest content analysis.

Recommendation engine boosted

Power your suggestions up by hyper-personalized recommendations with meta data derived from visuals. Make metadata granulation hundreds of times deeper by relying not only on textual descriptions but on a video itself.

Increased user satisfaction

Automated metadata enrichment works hundreds of times faster and shows better results by revealing hidden dependencies. All resulting in a retention rate boosted.

Plagiarism detection and anti-piracy analysis

Analyze on a scene level

The system analyses video content on a scene level rather than comparing separate frames. This ensures full contextual understanding of whether plagiarism is present.

Fight digital piracy and hidden plagiarism

Leveraging genetic algorithms for similarity analysis, the system understands which parts of content are stolen or copied contextually. By that you can ensure that your content is unique and will not be appealed against.

Exclude improper content from your broadcasting

Secure your brand by excluding illegal content from your broadcast once you’ve found it. Make sure your brand is safe from online piracy.

Video clustering

Frame-precise video segmentation

The system analyses video content on a scene level rather than looking at separate frames. It splits video timeline in parts logically, understanding the context of a scene as a whole. Yet it doesn’t leave any of the frames behind.

Avoid scene splitting

Seamless experience with video content is easy to achieve. No interruption by an untimely ad block in the middle of a scene. Know that nothing will spoil intensity of emotions.

Know your content composition

Unsupervised learning techniques tackled by us ensure that the system gets your content split into logically-divided parts automatically without any effort from your side. Where each video cluster represents a certain semantical part of the content.

Content compliance proof

Mark unwanted content

Find all the types of content that might not be accepted by your viewers or regulations (like nudity, violence, smoking, alcohol, or any specific scene of interest).

Scene-based approach

It’s not always obvious what’s happening on a frame without context. That’s why we look at a whole scene rather than at separate pictures to get a full semantic understanding of sensitive content.

Ensure full compliance automatically

Make sure you offer content which is compliant with all the rules of the different geographical regions. Automate all the costly operations associated with content compliance.

Unsupervised abnormal content (behaviour) analysis

Detect anything outstanding

See if something in your broadcast is going not the way it used to. Let the system do the hard job and a human to make a decision over it.

No additional training needed

There’s no need to train the system to recognize all the unexpected events. The system just knows what content is ordinary and when something different happens.

Automatically alerting the operator

Get instant alerts when your usual stream is interrupted. Let your operator know in a blink of an eye that something needs their attention.