Skip to content

3. Meeting Notes 09|27

Daniel Obeng edited this page Nov 8, 2017 · 1 revision

COMMENTS BY PROFESSOR

Make sure audio and video work together side by side

Remember, if you just have one picture, you neglect so much other information. You don't have to make it too hard, you have to set some limits. You can ask specific questions that work for your program and no one will blame you.

If it's hard, don't go for it.

In the beginning, we can limit to a small data set.

There are open source software of analyzing sound, we can use existing stuff.

Stick to stuff that is already existing and has a lot of support.

RESOLUTION

  1. Lie detector system using video and Audio
    • input: video of a person talking

    • output: instances where the person is lying

    • Research and data is available

    • Data Needed:

      • How to determine a person is lying from facial expression
      • How to determine a person is lying from their voice
  • Very interesting : but could be very hard because lie detection is very hard
  • but what kind of lie do you want to detect
  • Specific questions and see how to people react to it
  • Then from that point see how they react to it

Scenarios: Airport or traffic stop etc.

  1. Threat detection system using video and Audio
    • He thinks threat is easier
    • you can focus on a small threat, whether or not someone wants to beat you or not.

Clone this wiki locally