US 12,452,540 B1
Object-based auto exposure using neural network models
Chih-Chun Lee, Hsinchu (TW); Pei-Chen Lin, Tainan (TW); and Pei-Chien Yu, Taipei (TW)
Assigned to Ambarella International LP, Santa Clara, CA (US)
Filed by Ambarella International LP, Santa Clara, CA (US)
Filed on Sep. 13, 2022, as Appl. No. 17/943,472.
Int. Cl. H04N 23/73 (2023.01); G06T 7/50 (2017.01); G06T 7/60 (2017.01); G06T 7/73 (2017.01); G06V 10/82 (2022.01); G06V 20/40 (2022.01); G06V 20/56 (2022.01); H04N 23/71 (2023.01)
CPC H04N 23/73 (2023.01) [G06T 7/50 (2017.01); G06T 7/60 (2013.01); G06T 7/74 (2017.01); G06V 10/82 (2022.01); G06V 20/40 (2022.01); G06V 20/56 (2022.01); H04N 23/71 (2023.01); G06T 2207/10016 (2013.01); G06T 2207/20021 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30252 (2013.01)] 19 Claims
OG exemplary drawing
 
1. An apparatus comprising:
an interface configured to receive pixel data; and
a processor configured to (i) process said pixel data arranged as video frames, (ii) perform computer vision operations on said video frames to detect objects in said video frames, (iii) analyze characteristics of said objects detected in a current video frame of said video frames, (iv) determine adaptive auto-exposure weightings for said current video frame in response to said characteristics of said objects detected, (v) generate an auto-exposure weighting table based on said adaptive auto-exposure weightings specific to said current video frame and (vi) generate output video frames, wherein
(a) each of said output video frames is generated in response to using said auto-exposure weighting table specific to said current video frame to adjust luma values in said current one of said video frames,
(b) said auto-exposure weighting table is individually generated for only specific use with said current one of said video frames,
(c) said processor is configured to generate each of said output video frames in response to (A) using said auto-exposure weighting table to adjust said luma values in said current video frame when one of said objects is detected in said current video frame based on increasing default weight values in response to said characteristics of said objects and (B) using a fixed weighting table to adjust said luma values in said current video frame when none of said objects are detected in said current video frame, and
(d) said fixed weighting table comprises pre-defined luma adjustment values selected for a particular type of scene captured in said video frames from a plurality of predefined scene types.