Collection of machine learning training data for expression recognition

A non-transitory computer-readable medium storing instructions executable by a processor, the instructions comprising instructions to:

• receive, responsive to one or more requests for one or more images depicting a specific affective state, image data corresponding to at least one image from one or more first devices, wherein the affective state is one of anger, contempt, disgust, fear, happiness, sadness, and surprise;
• receive, from one or more second devices, one or more selections corresponding to the image data;
• determine one or more ratings relating to the image data, each of the one or more ratings corresponding to at least one selection from amongst the one or more received selections;
• test a machine learning classifier by:
• presenting, based at least in part on the one or more ratings, the image data to the machine learning classifier;
• receiving, from the machine learning classifier, a classification corresponding to the image data;
• and
• detecting whether the classification corresponds to the specific affective state;
• and
• provide, responsive to the detection, training data to the machine learning classifier.

A computer system comprising;

• at least one processor;
• at least one memory, coupled to the processor and storing instructions, the instructions comprising instructions that when executed by the processor, cause the processor to:
• receive, responsive to one or more requests for one or more images depicting a specific affective state, image data corresponding to at least one image from one or more first devices, wherein the affective state is one of anger, contempt, disgust, fear, happiness, sadness, and surprise;
• receive, from one or more second devices, one or more selections corresponding to the image data;
• determine one or more ratings relating to the image data, each of the one or more ratings corresponding to at least one selection from amongst the one or more received selections;
• present, based at least in part on the one or more ratings, the image data to the machine learning classifier;
• receive, from the machine learning classifier, a classification corresponding to the image data;
• detect whether the classification corresponds to the specific affective state;
• and
• provide, responsive to the detection, training data to the machine learning classifier.