Automatically generating best digital images of a person in a physical environment

A system for generating best images of a person in a retail environment, the system comprising:

• a camera positioned within a retail environment, the camera comprising at least one image sensor that is configured to generate a continuous stream of image data of the retail environment;
• and
• an edge computing device in communication with the camera, the edge computing device configured to:
• receive, from the camera, the continuous stream of image data;
• detect, using object detection techniques, a person in the continuous stream of image data, wherein the continuous stream of image data comprises a plurality of images that are part of a time series;
• generate bounding boxes for each of the plurality of images around the person based on detecting the person as the person moves in the continuous stream of image data;
• identify, based on applying a features model to each of the bounding boxes, at least one feature of the plurality of images depicting the person, the features model having been trained to identify and score features in training image data, the training image data comprising at least one of facial features of a person and quality features of the training image data;
• select a subset of the bounding boxes having at least one of the at least one features that satisfies best images criteria, wherein the subset of the bounding box images are best images of the person;
• and
• return the best images of the person.