Video processing system and method for object detection in a sequence of image frames

A method for image processing, comprising:

• receiving, by circuitry in an electronic device, a first object template for a first object in a first image frame, wherein
• said first image frame comprises said first object and a second object, and
• said first object is at a first distance from said second object in said first image frame;
• determining, by said circuitry, a plurality of object candidates corresponding to said first object,
• wherein said plurality of object candidates are determined for a second image frame based on a shape of said first object template, wherein
• said first object is at a second distance from said second object in said second image frame, and
• said second image frame is subsequent to said first image frame in a sequence of image frames;
• selecting, by said circuitry, a second object template from said determined plurality of object candidates based on at least one first object parameter, said first distance, and said second distance, wherein
• said at least one first object parameter corresponds to a third distance of a plurality of object center coordinates of said plurality of object candidates in said second image frame, with respect to a center coordinate of said first object template in said first image frame, and
• said center coordinate is different from said plurality of object center coordinates;
• updating, by said circuitry, said first object template to said second object template;
• and
• segmenting, by said circuitry, said first object in said second image frame based on said second object template.

A non-transitory computer-readable medium having stored thereon, computer-executed instructions, which, when executed by a processor of a video-processing device, cause said processor to execute operations, said operations comprising:

• receiving a first object template for a first object in a first image frame, wherein
• said first image frame comprises said first object and a second object, and
• said first object is at a first distance from said second object in said first image frame;
• determining a plurality of object candidates corresponding to said first object, wherein
• said plurality of object candidates are determined for a second image frame based on a shape of said first object template,
• said first object is at a second distance from said second object in said second image frame, and
• said second image frame is subsequent to said first image frame in a sequence of image frames;
• selecting a second object template from said plurality of object candidates based on at least one first object parameter, said first distance, and said second distance, wherein
• said at least one first object parameter corresponds to a third distance of a plurality of object center coordinates of said plurality of object candidates in said second image frame, with respect to a center coordinate of said first object template in said first image frame, and
• said center coordinate is different from said plurality of object center coordinates;
• updating said received first object template to said second object template;
• and
• segmenting said first object in said second image frame based on said second object template.