Method and system for training a machine learning model with a subclass of one or more predefined classes of visual objects

A system for training a machine learning model with a subclass of one or more predefined classes of visual objects obtained from one or more videos, the system comprising:

• a computer memory configured to store one or more input videos comprising visual objects;
• a classifier implemented by a computer processor configured to classify the visual objects from the input videos into a plurality of predefined classes;
• an electronic display configured to present to a human operator, a plurality of presented visual objects obtained from the one or more videos, belonging to one of the one predefined class requested by the human operator;
• a user interface associated with the electronic display, configured to receive from the human operator, selected visual objects being a selection of some of the plurality of the presented visual objects, wherein the selection is directed at visual objects belonging to at least one subclass of the one predefined class and
• a machine learning module, implemented by the computer processor, configured to create additional samples of the selected visual objects by tracking said selected visual objects across video frames of the one or more videos and to carry out supervised training of a classifier, using the selected visual objects, the additional samples of the selected visual objects, and non-selected visual objects as negative examples, to distinguish between the selected visual objects and the presented visual objects which were not selected,
• wherein the computer processor applies the classifier as a subclass filter for classifying visual objects in other videos, as belonging to the selected at least one subclass.

A non-transitory computer readable medium for training a machine learning model with a subclass of one or more predefined classes of visual objects obtained from one or more videos, the computer readable medium comprising a set of instructions that, when executed, cause at least one computer processor to:

• classify the visual objects from the one or more videos into a plurality of predefined classes;
• present to a human operator, over an electronic display, a plurality of presented visual objects obtained from the one or more videos, belonging to the one predefined class requested by the human operator;
• receive from the human operator, over a user interface associated with the electronic display, a selection of some of the plurality of visual objects, wherein the selection is directed at visual objects belonging to at least one subclass of the one predefined class;
• create additional samples of the selected visual objects by tracking said selected visual objects across video frames of the one or more videos and extracting corresponding tracked thumbnail images of the selected visual objects from different frames;
• carry out supervised training of a classifier using the selected visual objects, the additional samples of the selected visual objects, and non-selected visual objects as negative examples, to distinguish between the selected visual objects and the presented visual objects which were not selected;
• and
• apply the classifier as a subclass filter for classifying visual objects in other videos, as belonging to the selected at least one subclass.