Method, apparatus, and system for automatically annotating a target object in images

A method comprising:

• obtaining an image training sample comprising a series of images, each image of the series of images obtained by photographing a target object, and adjacent images of the series of images sharing one or more environmental feature points;
• using one of the series of images as a reference image to determine a reference coordinate system, and generating a three-dimensional space model based on the three-dimensional reference coordinate system;
• determining, when the three-dimensional space model is moved to a position of the target object in the reference image, position information of the target object in the three-dimensional reference coordinate system;
• and
• mapping the three-dimensional space model to image planes of each image in the series of images based on the position information of the target object in the three-dimensional reference coordinate system and respective camera pose information corresponding to each image in the series of images, the respective camera pose determined based on environmental feature points in each image.

A non-transitory computer-readable storage medium for tangibly storing computer program instructions capable of being executed by a computer processor, the computer program instructions defining the steps of:

• obtaining an image training sample comprising a series of images, each image of the series of images obtained by photographing a target object, and adjacent images of the series of images sharing one or more environmental feature points;
• using one of the series of images as a reference image to determine a reference coordinate system, and generating a three-dimensional space model based on the three-dimensional reference coordinate system;
• determining, when the three-dimensional space model is moved to a position of the target object in the reference image, position information of the target object in the three-dimensional reference coordinate system;
• and
• mapping the three-dimensional space model to image planes of each image in the series of images based on the position information of the target object in the three-dimensional reference coordinate system and respective camera pose information corresponding to each image in the series of images, the respective camera pose determined based on environmental feature points in each image.

An apparatus comprising:

• a processor;
• and
• a storage medium for tangibly storing thereon program logic for execution by the processor, the stored program logic comprising:
• logic, executed by the processor, for obtaining an image training sample comprising a series of images, each image of the series of images obtained by photographing a target object, and adjacent images of the series of images sharing one or more environmental feature points,
• logic, executed by the processor, for using one of the series of images as a reference image to determine a reference coordinate system, and generating a three-dimensional space model based on the three-dimensional reference coordinate system,
• logic, executed by the processor, for determining, when the three-dimensional space model is moved to a position of the target object in the reference image, position information of the target object in the three-dimensional reference coordinate system, and
• logic, executed by the processor, for mapping the three-dimensional space model to image planes of each image in the series of images based on the position information of the target object in the three-dimensional reference coordinate system and respective camera pose information corresponding to each image in the series of images, the respective camera pose determined based on environmental feature points in each image.