| CPC G06V 40/28 (2022.01) [G06V 10/776 (2022.01); G06V 40/176 (2022.01)] | 20 Claims |

|
1. A method, comprising:
capturing, using a camera, a video containing a plurality of sign language motions made by a first individual;
predicting, using a trained machine learning (ML) model, a first mood of the first individual based on characterizing the plurality of sign language motions in the video;
predicting, using a first sentiment analysis technique, a second mood of the first individual based on other information in the video besides the plurality of sign language motions;
determining there is a conflict between the first mood predicted using the trained ML model and the second mood predicted using the first sentiment analysis technique; and
retraining, after determining the conflict, the ML model based on an annotated video of the first individual performing a plurality of sign language motions.
|