FaceCrop™

Video Compression powered by facial recognition

 Problem

Video delivery on

1) Poor bandwidths

2) Small screen devices (old phones and smart watches)

 End-Product-Focused Solution

Content discovery platform,

with dynamic video caching,

that works on low bandwidth networks

as well as offline

(android and iphone app)

Who Uses or Benefits From the End-Product Focused Solution 

People who do live in areas of poor internet penetration.

People who spend a lot of time offline.

• Splitting Video in frames

• Facial Recognition

• Generate face frames

• Compiling video from face frames

Solution

200kb

1mb

Original Video

FaceCrop™ed video

Solution : Sample

2.5mb

500kb

Solution : Sample

Original Video

FaceCrop™ed video

8mb

2mb

Original Video

FaceCrop™ed video

Solution : Sample

python facecrop.py    <youtube-url>

python facecrop.py    path/to/video.mp4

Use Chats Bots to find Content on our platform

Use our Portal to upload new content

Solution : Technical details

Content consumption medium around educational videos

p2p video conferencing app the works best on low bandwidth

Exploring use cases in other domains like drone video compression.

Product timeline

June 2016

October 2016

December 2016

Improvements: Video stabalisation, noise removal

April 2016

Original video 9MB

Bugdet

$40,000

only

Predicted Hurdles

Video Stabilization

iPhone/Android App

3D face modelling

p2p video conf

Exploring Use Cases

3D weather data compression

Exploring Use Cases

Relaying Video on FM bandwidth

Exploring Use Cases

P2P video chatting 

Team

Saurav Tomar

Pursuing bachelors degree from DTU (India)

Full stack hacker

Currently running a video streaming startup Chestream.com

Will work on core software development and hiring.

Jumana Karwa

Pursuing Masters Degree at UW

10 yrs of java dev experience

Profound project manager experience

Will work on business development and project management

FaceCrop