Overview

[Joe24424]

Using a selection of body parameters eg angle that body is facing, movements of arms, shoulder width, and facial expression, to dictate what emotional state that a person is in. Identification of these emotional states. Creation of a repository (or whatever it is called) of body emotional state transformations. Creation of a vocabulary list with aggression markers. input of language causes output in body language and language. Kaggle.com is a useful repository of data

[Joe24424]

So the machine learning program runs something like this: There is a metric to determine emotional state, and this needs to be very accurate, better than 95%. From this metric of determining emotional state, you extrapolate, and you say phrases that are positively weighted with positive emotional states, cause positive body movements, and positive discussion. You also need a discussion phrases, which will find the most suitable set of phrases, this can be from a very diverse dataset, and it will create the capacity to generate an intelligent response. Whether an intelligent response is generated or not will be dependent upon the SENTIMENT of the comments which was determined earlier. As camgirls don't talk much, and only talk when you make them happy, so should the model. Creating a facial moving model combined with talking animation. I can probably get taught how to do this by some animation video on you-tube.

So here is a general summary: 1>You create a happy/sad/nervous audio determinant and video determinant 2>You parse livecams and weight text as happy/sad/nervous inducing text 3>You parse livecams and extract extreme point movements, to control facial expressions, and talking. 4>You create bones using blender tool (you-tube search it), and create a moving model 5>Since innevitably the bones are going to be a difficulty fuck fuck fuck 4>You test 5>You apply changes etc. 6>wahoo you are no-longer dead.

[Joe24424]

1>Under 2 in camgirl we can make a parser and create a place for the dataset which is created to be placed in 2/data. You can find examples of python parsers in the python index I need to find examples of audio where someone is happy or sad etc

[Joe24424]

2>The happy/sad/nervous audio/video determinant datasets: Read some python book for this, and while it is still present in mind, finish this portion of the project, extremely important, since you have been reading stuff along these lines for a while, and still only have a listening comprehension of things. The audio is attained from small sections of livecams having a whinge (which I will extract since you don't need excessive audio. You have a folder on your computer under projects,camgirl called 1 that you can open. This folder needs to have a folder data in it. In the data folder we will place both the audio, and the video determinant. We will also publish a link to the kaggle kernel.

[Joe24424]

3> Save all your point recognition stuff under camgirl/3 yes sir you need to be organized. yes sir yes sir.

[Joe24424]

4> Simple as is, save the model with the bones under 4/model