Connecting relevant video content to audiences CREDENTIALS DECK 1 - - PowerPoint PPT Presentation

▶
connecting relevant video content
SMART_READER_LITE
LIVE PREVIEW

Connecting relevant video content to audiences CREDENTIALS DECK 1 - - PowerPoint PPT Presentation

Oct 06, 2023 36 likes •206 views

Connecting relevant video content to audiences CREDENTIALS DECK 1 Hello, Were Vilynx We've developed machine learning technology to drive video discovery and engagement . US company , headquarters in Palo Alto, offices in New York and

slide-1
SLIDE 1

1

Connecting relevant video content to audiences

CREDENTIALS DECK

slide-2
SLIDE 2

2

We've developed machine learning technology to drive

video discovery and engagement.

US company, headquarters in Palo Alto, offices in New York

and Barcelona (Spain).

Top ML/DL team with more than 200 patents among

everyone

Funding: $2.6M seed Capital.

Hello, We’re Vilynx

slide-3
SLIDE 3

3

Millions hours of video are recorded every sec. Millions hours of video are uploaded to the cloud every

sec.

Video creation growing exponentially due to live feeds.

Why?

slide-4
SLIDE 4

4

  • Visually: Static picture/thumbnail.
  • A static image cannot describe the most relevant moment of

the video.

  • Search & Recommendation engine: key words of the

video/metadata & audience preference

  • No metadata from trending conversations in social media/web.
  • No metadata associated to relevant parts inside the video.
  • Most metadata currently added manually – doesn’t scale

How any of us can navigate through all those hours of video?

slide-5
SLIDE 5
  • Previews - Real time selection
  • f the best moments of a video.
  • Metadata - Real time auto

tagging those moments with social media/web key words.

  • Personalization - Real time

selection based on each user preferences.

5

How do we improve video discoverability?

slide-6
SLIDE 6
  • Deep Networks today: match images to tags/key words
  • Images are not “moments/actions”.

6

What does Machine Learning & Deep Learning do today?

slide-7
SLIDE 7

Current training sets are based on public databases of images and the keyword/ tags are manually selected. Those key words don’t describe what happens in real time. Live sporting events, terrorist attacks, concerts….. People don’t search for ‘tennis court’ or ‘tennis player’, people look for ‘Wimbledon 2016’ and ‘Andy Murray’. Auto-tagging video requires training datasets based on key words people use to describe those videos, moments, actions.

7

How does ML/DL auto-tagging work today?

slide-8
SLIDE 8

No they don’t. How can we train hours of video in real time? Too much GPU/CPU processing Real time content requires a new architecture for ML/DL.

8

Do current solutions learn from videos and social media in real time today?

slide-9
SLIDE 9

How much info is relevant within a video for a Deep Network? The inverse Autocorrelation function show that most of time is zero, no new info but ….. There are the key moments

9

How to train Deep Networks with video in realtime?

slide-10
SLIDE 10

10

A data training set of video moments that will include:

  • Set of frames that describes it: 5 sec clips.
  • Key words that describes that moment, including social

media.

Update the training set in real time with new videos and

new key words.

Train the network in real time to be able to auto-tag new

videos in real time.

How to move our Deep Learning Networks from images to “video moments”?

slide-11
SLIDE 11

11

We are creating a data set for DL training with 5 to 7 sec clips

  • f videos and key words from social media / web

Lionel Messi Amazing Second Goal Barcelona vs Bayern Munich score 2 - 0 May 06 2015 Champions League James Harden final basket Rockets versus New York Nets

slide-12
SLIDE 12

12

Here’s how it works.

  • We ingest customer videos

and the contextual information around it.

  • We then take cues from

around the Web and social networks.

  • This combined input is fed

to the most advanced convolutional deep neural network in the industry.

  • Output are video previews
  • ptimized to engage your

audience and rich metadata that can further drive your video content.

slide-13
SLIDE 13

13

DL Through a Distributed Multi Convolutional Neural Networks

  • Distributed network architecture
  • Multiple Networks rather than a

single huge network

  • Real Time Training
  • 80% accuracy on selecting frames

per video Taking advantage of all current high-performance HW, servers/GPUs

slide-14
SLIDE 14

Use Video Processing Algorithms to find the relevant parts of the videos, taking into account:

  • Image recognition (objects, people, places)
  • Movement, colors, diversity
  • Audience preferences
  • Trending topics

Algorithms trained using ML (audience behavior analytics, individual scores per groups of videos, etc..)

14

Video 5-Sec Previews

More than 400k videos processed 300M summaries served 1.2B audience data points. 98% accuracy to find the relevant parts of the video CTR increase between 50% to 500% (customer validated)

slide-15
SLIDE 15

DL algorithms to extract video keywords Our training DB learns from any online video and ads this information to the keyword selection Reaching to 90% top-1 accuracy on keyword prediction

  • Title
  • Meta-Keywords
  • Social Trends
  • Full text
  • Video URL

15

Video Metadata

slide-16
SLIDE 16

Huge database of tagged videos

* 400k videos with * 20k unique tags clustered into * 100 different clusters.

16

DL Network Video Training Data Set

This results in 50M tagged images, 2.5M of video moments/actions And growing every day

slide-17
SLIDE 17

17

Thank You

22

Juan Carlos Riveiro Chief Executive Officer jc@vilynx.com Palo Alto • New York City • Barcelona