US 12,347,120 B2
Vision-based target tracking method, system, equipment, and storage medium
Ming Zhang, Guangdong (CN); Jian Dong, Guangdong (CN); and Shuai Li, Guangdong (CN)
Assigned to REMO TECH Co., Ltd., Shenzhen (CN)
Filed by REMO TECH Co., Ltd., Guangdong (CN)
Filed on Aug. 12, 2022, as Appl. No. 17/886,515.
Application 17/886,515 is a continuation of application No. PCT/CN2020/076360, filed on Feb. 24, 2020.
Claims priority of application No. 202010095228.9 (CN), filed on Feb. 14, 2020.
Prior Publication US 2022/0383518 A1, Dec. 1, 2022
Int. Cl. G06T 7/246 (2017.01); G06T 7/73 (2017.01); G06V 10/74 (2022.01); G06V 40/16 (2022.01)
CPC G06T 7/248 (2017.01) [G06T 7/73 (2017.01); G06V 10/761 (2022.01); G06V 40/168 (2022.01); G06T 2207/30201 (2013.01)] 13 Claims
OG exemplary drawing
 
1. A non-transitory computer-readable medium with instruction stored thereon, that when executed by a processor, perform vision-based target tracking method comprising steps:
step A: reading a current image frame so as to acquire and save a face position and a body position of all persons in the current image frame;
step B: acquiring a previous image frame and a position of a to-be-tracked target in previous frame, and acquiring, according to the current image frame, the previous image frame and the position of the target in previous image frame, a current position of a time-series tracking target by using a time-series tracking algorithm;
step C: determining whether a person matching a face of the to-be-tracked target exists outside a region corresponding to the current position of the time-series tracking target;
step D: when there is no a person matching a face of the to-be-tracked target existing outside a region corresponding to the current position of the time-series tracking target, determining whether the time-series tracking target is in an intersecting state according to the current position of the time-series tracking target and the body position of other persons;
step E: when the time-series tracking target is in an intersecting state, determining whether a person matching a body of the to-be-tracked target exists amongst the other persons intersecting with the time-series tracking target, when there is no a person matching a body of the to-be-tracked target exists amongst the other persons intersecting with the time-series tracking target, taking the current position of the time-series tracking target as a current target position;
step F: keeping tracking; and
step G: taking the current target position as the position of the to-be-tracked target in the previous image frame, taking the current image frame as the previous image frame, the time-series tracking algorithm, face recognition and body recognition are combined to obtain the current target position as the position of the to-be-tracked target for target tracking by reducing possibility of a sudden change in the posture of the target person, and reducing the tracking error during irregular intersection of the target person and other persons, and the face recognition is set as the first priority of target switching.