Computing device and method using a neural network to determine whether or not to process images of an image flow

A method using a neural network to determine whether or not to process images of an image flow, the method comprising:

• (a) storing in a memory of a computing device a predictive model generated by a neural network training engine;
• (b) receiving by a processing unit of the computing device an image of the image flow;
• (c) executing by the processing unit of the computing device a neural network inference engine implementing the neural network, the neural network using the predictive model for generating an indication of whether or not to process the image based on one or more input of the neural network, the one or more input comprising the image and a previous indication of whether or not to process a previous image of the image flow;
• (d) determining by the processing unit of the computing device whether or not to process the image by an image processing module executed by the processing unit of the computing device based on the indication of whether or not to process the image;
• (e) processing the image by the image processing module if the determination is positive and not processing the image by the image processing module if the determination is negative;
• and
• (f) storing the indication of whether or not to process the image in the memory of the computing device, the indication being used as input of the neural network along with a following image of the image flow.

A method for training a neural network to determine whether or not to process images of an image flow, the method comprising:

• (a) initializing by a processing unit of a computing device a predictive model of a neural network;
• (b) generating by the processing unit of the computing device training data, the generation of the training data comprising:
• generating a plurality of image flows, each image flow comprising a plurality of images;
• and
• determining for each image of the plurality of image flows a target indication, the target indication indicating whether or not to process the image by an image processing module;
• (c) executing by the processing unit of the computing device a neural network training engine implementing the neural network, the neural network using the predictive model for generating an inferred indication of whether or not to process a given image of a given image flow among the plurality of image flows based on one or more input, the one or more input comprising the given image of the given image flow;
• and
• (d) adjusting by the processing unit of the computing device the predictive model of the neural network to minimize a difference between the inferred indication and the target indication determined for the given image of the given image flow;
• wherein determining the target indication for the given image of the given image flow among the plurality of image flows comprises:
• determining a number of persons present in the given image;
• comparing the number of persons present in the given image with a previously determined number of persons present in a previous image of the given image flow;
• and
• if the compared numbers of persons are different then setting the target indication to a value indicating to process the given image else setting the target indication to a value indicating not to process the given image.