US 12,073,602 B2
Automated key frame selection
Chun-Hao Liu, Fremont, CA (US); Jayanta Kumar Dutta, San Jose, CA (US); and Naveen Ramakrishnan, Campbell, CA (US)
Assigned to Robert Bosch GmbH, Stuttgart (DE)
Filed by Robert Bosch GmbH, Stuttgart (DE)
Filed on Feb. 17, 2022, as Appl. No. 17/674,230.
Prior Publication US 2023/0260251 A1, Aug. 17, 2023
Int. Cl. G06V 10/00 (2022.01); G06V 10/762 (2022.01); G06V 10/77 (2022.01)
CPC G06V 10/762 (2022.01) [G06V 10/7715 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A method for identifying key frames of a video, comprising:
performing object detection to identify frames of a video including target classes of objects of interest;
extracting features of the identified frames to generate raw feature vectors;
compressing the feature vectors into lower dimension vectors;
clustering the compressed feature vectors into a plurality of clusters;
filtering the clustered compressed feature vectors to identify the key frames from each of the plurality of clusters, including selecting the key frames by including one key frame from each cluster determined as having a median detection score for a target class of the target classes of objects of interest; and
providing the key frames as a representative data set of the video.