AFFECTIVE COMPUTING

Problems regarding accuracy and functionality

 

- As mentioned in the modelling section, higher level of details comes with the cost of extreme increase in calculation  time and intensity.

 

- 2 or more emotions can have similar gestures with only minor differences makes it difficult to accurately recognise  an emotion.

 

- To be able to detect a wide range of emotions, the data collection has to be massive, thus the running time will  increase due to longer patterm matching of the input and database.

 

- For 3D models analysis, small amount of cameras used will possibly end up with hidden spots, resulting in an  incomplete model for analysis. Otherwise the cost will be high making the detection not practical for common usage.

 

- Input for the same emotion for each person can almost never be identical as interpretation for each emotion may  vary. This also means one action for different people may convey different emotions.

 

-Unconsistent lighting and change in environment can cause image noise for 2D capturing and use of RGB cameras.      This lowers the accuracy for gesture recognition.