US 11,657,123 B2
Method and apparatus for people flow analysis using similar-image search
Yuki Watanabe, San Jose, CA (US); and Ravigopal Vennelakanti, San Jose, CA (US)
Assigned to HITACHI, LTD., Tokyo (JP)
Filed by Hitachi, Ltd., Tokyo (JP)
Filed on Oct. 8, 2020, as Appl. No. 17/66,121.
Prior Publication US 2022/0114381 A1, Apr. 14, 2022
Int. Cl. G06K 9/62 (2022.01); G06F 18/22 (2023.01); G06V 40/16 (2022.01); G06V 10/75 (2022.01)
CPC G06F 18/22 (2023.01) [G06V 10/7515 (2022.01); G06V 40/169 (2022.01); G06V 40/172 (2022.01)] 12 Claims
OG exemplary drawing
 
1. A method, comprising:
detecting and tracking a person from images from a first camera;
extracting image features from the detected and tracked person from the images from the first camera;
storing the image features with information associated with the camera into a database;
conducting a sample query to search the database for another person detected from images from a second camera;
for the another person being within a threshold of similarity of the person from the images of the first camera;
determining a gap of time between the person from the images of the first camera and the another person from the images from the second camera; and
calculating travel time for the person from the gap of time,
wherein the detecting and the tracking the person from the images from the first camera further comprises detecting and tracking one or more other persons in proximity to the person, and
wherein the extracting the image features from the detected and tracked person from the images from the first camera further comprises extracting the image features from the one or more other persons in proximity to the person to derive group features.