Generating highlights for live videos such as sports match, presidential debates etc. needs experts and also viewers might have to wait for some time in order to watch those highlights.
We present LiveLight which uses every viewer's individual moments (moments they love) to generate highlights during the live stream.
As far as we know, our proposed system is first of its kind that uses crowdsourcing for highlight generation. The unique feature of our system is that it processes crowd's moments during the live stream and generates highlights before the live stream ends.
At first all user must be signed up to use our system.
After logged in, user can use our system.
Livelight is implemented using node.js v6.6.0 as a serverside framework and bootstrap Bootstrap v3.3.7 as a front-end web page development.
To make the system look better, we used Bootflat v2.0.4 css skins along with BootStrap.
The interaction between the server and client is done two ways: 1) Ajax(Asynchronous JavaScript and XML), 2) WebSocket. In order to handle Ajax Restful APIs on front-end, we used JQuery v3.1.1. In order to handle socket-based communication, we used socket.io v1.4.5.
In order to show videos, we used video.js v4.3.0, and used extensions like videojs-markers v0.6.1 and videojs.disableProgress. Also, we did customization on the css to make it more suitable for our purpose.
For mongoDB ODM, we used mongoose v4.4.
Also, we generate thumbnail images with html2canvas v0.5.0-beta3.
Finally, the authentication of users is done with passport v0.1.18.
The server is running on Heroku(https://www.heroku.com/), which is a platform-as-a-service(PaaS). We used hobby($7) sized instance for our system. As Heroku nicely supports scaling the system, we can handle the scalability issue pretty easily.
mongoDB is running on mLab(https://mlab.com/), which is a mongoDB platform that hosts mongoDB.
On top of above infrastructure, we run our node.js express server and bootstrap & jQuery based front-end.
You can use our prototype here: https://live-light.herokuapp.com/
The repository of Livelight is available at https://gitlab.com/CS492/term_project.git