Image processing apparatus, image processing method, and storage medium

An image processing method comprising:

• acquiring an image of at least one person;
• detecting at least one person in the acquired image;
• controlling processing for tracking the at least one detected person;
• controlling processing for detecting an action of the at least one detected person based on a result of tracking the at least one detected person, and
• controlling processing for estimating position information about a component of the at least one detected person,
• wherein the result of tracking the at least one detected person includes state information representing a tracking state of the at least one detected person,
• wherein whether to use the result of tracking the at least one detected person is based on a type of the action detection processing and the state information,
• wherein the state information is updated to indicate a suspended state in a case where, among tracking likelihoods calculated for associating the at least one detected person with each of a plurality of previously detected persons, two or more of the tracking likelihoods are greater than or equal to a predetermined likelihood threshold,
• wherein, in a case where the type of the action detection processing is a type not based on time-series variation, the result of tracking the at least one detected person indicating the suspended state is used in executing the action detection processing, and in a case where the type of the action detection processing is a type based on time-series variation, the result of tracking the at least one detected person indicating the suspended state is not used in executing the action detection processing, and
• wherein the action detection processing of the type based on time-series variation includes detecting a behavior of the at least one detected person based on time-series variation of the position information about the component and detecting a surroundings check action of the at least one detected person based on a difference between a direction of the at least one detected person's face and a value predicted from the face direction.

A non-transitory computer-readable storage medium storing a program for causing a computer to execute an image processing method, the image processing method comprising:

• acquiring an image of at least one person;
• detecting at least one person in the acquired image;
• controlling processing for tracking the at least one detected person;
• controlling processing for detecting an action of the at least one detected person based on a result of tracking the at least one detected person, and
• controlling processing for estimating position information about a component of the at least one detected person,
• wherein the result of tracking the at least one detected person includes state information representing a tracking state of the at least one detected person,
• wherein whether to use the result of tracking the at least one detected person is based on a type of the action detection processing and the state information,
• wherein the state information is updated to indicate a suspended state in a case where, among tracking likelihoods calculated for associating the at least one detected person with each of a plurality of previously detected persons, two or more of the tracking likelihoods are greater than or equal to a predetermined likelihood threshold,
• wherein, in a case where the type of the action detection processing is a type not based on time-series variation, the result of tracking the at least one detected person indicating the suspended state is used in executing the action detection processing, and in a case where the type of the action detection processing is a type based on time-series variation, the result of tracking the at least one detected person indicating the suspended state is not used in executing the action detection processing, and
• wherein the action detection processing of the type based on time-series variation includes detecting a behavior of the at least one detected person based on time-series variation of the position information about the component and detecting a surroundings check action of the at least one detected person based on a difference between a direction of the at least one detected person's face and a value predicted from the face direction.