Gumapp

AI POWERED VIDEO ENRICHMENT

We build artificial intelligence solutions to maximize the value of video content for the media industry.

OUR PRODUCTS
go down
icon tag

SMART VIDEO THUMBNAILS

icon tag

AUTOMATIC VIDEO TAGGING

icon tag

IN-VIDEO SEARCH

Selecting the right thumbnails for a video is essential to engage viewers, but how to select the best image from a video that contains thousands of frames ? Manual selection is time consuming and might not end up with an optimal Click Through Rate (CTR) or a satisfying video completion rate.

Our solution automatically selects relevant thumbnails from a video and then uses machine learning techniques to find the best image that creates the maximum engagement. 

LEARN MORE

AUTOMATIC VIDEO TAGGING

As the volume of online video content available is growing faster than ever, the importance of having accurate and fine-grained metadata becomes crucial. Tagging video content manually is time consuming and frequently inconsistent : tags are missing, overlapping or not compliant with the taxonomy. A large video catalog might have been tagged over the time by different teams and would require to be re-tagged entirely, but tagging it manually is too expensive.

Our solution automatically generates tags for each piece of content of your catalog by analyzing and understanding what’s inside the video.

LEARN MORE

IN-VIDEO SEARCH

A video is a type of black box for search engines.
Search engines can’t search inside a video file like they do inside a text file. Instead they rely on metadata attached to each video asset, typically the video title, manually added tags and the description. Finding the right sequence in a video is not an easy task, especially when dealing with long- form videos such as TV programs. In addition video metadata is often incomplete and inaccurate and you might even not know inside which video you should be looking for. For instance you might be looking for an interview of your favorite singer in a TV program, but how to find it if the name of the singer doesn’t even appear in the program synopsis ? In a context where viewers shift massively to short-form videos, the ability to create or play selected clips from long-form videos becomes vital.

Our solution automatically indexes every second of a video by analyzing dialogues and images, then you can search inside a video and play directly the right sequence.

LEARN MORE 
Client Client Client
Client Client Client
Client Client Client

Our references

We work with broadcasters, content providers, TV operators and online publishers. Some of our references : TF1, Canal+, France TV, M6, arte, BFM, Viacom, Bouygues Telecom, Ouest France.

AWARDS


Ubifrance Silicon Valley Tech Tour finalist

NPA Awards : best catch up TV service for France TV 

Ouest France : member of the start-up accelerator 

SFR Jeunes Talents Start Up 

Grand Prix Internet ESSEC Sénat

TECHNOLOGY


We use a blend of state-of-the art open source IA technologies and proprietary solutions. We have a proven experience in selecting the best technologies, combining them together and developping the missing parts to create efficient and flawless solutions.

NLP : speech-to-text, topic detection, named entity detection, sentiment analysis 

Computer vision : face recognition, object detection, OCR 

Machine learning : custom classifiers, neural networks, reinforcement learning

Research labs : CNRS, IRISA, Telecom Paris Tech

Drop us a line

You are very important to us, all information received will always remain confidential.