Cognition assistance

The system as claimed in claim I, wherein the video image frame is associated with audio data including voice data and non-voice data, wherein the input data receiver operates further to:

• determine the voice data from the audio data. based on predefined voice recognition models;
• annotate the determined voice data based on the predefined tags associated with the historical data;
• recognize a person related to the determined voice data based on a corresponding tag associated with the historical data;
• and
• convert the voice data into text data based on predefined speech-to-text models.

A computer-implemented method comprising:

• receiving preprocessed data by a processor in communication with an input data receiver coupled to a camera, the input data receiver operating to receive a video image frame captured by the camera, wherein the video image frame is related to an event, and preprocess the video image frame to extract facial data, scene data, and activity data therefrom in association with temporal data related thereto and geographical location data of the camera to provide the preprocessed data, wherein the scene data, the activity data, the geographical location data, and the temporal data collectively define spatiotemporal data;
• and
• performing, by the processor, training operations comprising:
• identifying an unknown aspect related to the preprocessed data based on a comparison with historical data, the unknown aspect being one of a face related to the facial data, a place related to the geographical location data, and a scene related to the scene data, wherein the unknown aspect is unassociated with predefined tags in the historical data;
• determining a predefined priority factor for the identified unknown aspect, wherein the predefined priority factor includes:
• a frequency of occurrence of the identified unknown aspect within a predefined period,
• a geographical proximity of the identified unknown aspect from the camera,
• a geographical proximity of the identified unknown aspect to a known face related to the facial data, wherein the known face is tagged in the historical data, and
• a geographical proximity of the identified unknown aspect to a known place tagged in the historical data;
• prioritizing the identified unknown aspect for annotation based on a value of the predefined priority factor being above a predefined threshold;
• annotating the prioritized aspect using a predetermined contextual tag;
• based on the annotated aspect, associating the facial data with the spatiotemporal data to provide contextual annotated data;
• and
• training a plurality of temporal models based on the contextual annotated data, wherein the plurality of trained temporal models and the contextual annotated data are used to provide an output corresponding to a mental memory of a user.