US 11,669,719 B2
Storage system of DNN outputs for black box
Jeremie Dreyfuss, Tel-Aviv (IL); Amit Bleiweiss, Yad Binyamin (IL); Lev Faivishevsky, Kfar Saba (IL); Tomer Bar-On, Petach-Tikva (IL); Yaniv Fais, Tal Aviv (IL); Jacob Subag, Kiryat haim (IL); Eran Ben-Avi, Haifa (IL); Neta Zmora, Tzur Moshe (IL); and Tomer Schwartz, Even Yehuda (IL)
Assigned to INTEL CORPORATION, Santa Clara, CA (US)
Filed by Intel Corporation, Santa Clara, CA (US)
Filed on Feb. 12, 2021, as Appl. No. 17/174,864.
Application 17/174,864 is a division of application No. 15/499,889, filed on Apr. 28, 2017, granted, now 10,922,556.
Prior Publication US 2021/0256272 A1, Aug. 19, 2021
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/063 (2023.01); G06N 3/08 (2023.01); G06N 3/04 (2023.01); G06N 3/084 (2023.01); G06V 20/56 (2022.01); G06V 10/44 (2022.01); G06F 18/214 (2023.01); G06N 3/044 (2023.01); G06N 3/045 (2023.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06F 18/2413 (2023.01)
CPC G06N 3/063 (2013.01) [G06F 18/214 (2023.01); G06F 18/24133 (2023.01); G06N 3/04 (2013.01); G06N 3/044 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06N 3/084 (2013.01); G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 20/56 (2022.01)] 23 Claims
OG exemplary drawing
 
1. An apparatus comprising:
a processor to:
analyze one or more backpropagated gradient maps from low resolution image data collected by a low resolution camera;
identify, from the low resolution image data and based on analysis of the one or more backpropagated gradient maps, at least one region of interest;
collect high resolution images from the at least one region of interest using a high resolution camera;
constrain a dropout in a neural network layer of a neural network model in order to remove at least one feature from an input to the neural network layer; and
upload at least a portion of the low resolution image data and the high resolution images to a datacenter for inclusion in the neural network model, wherein an approximation of an original image is re-created, based on the neural network model, using a pre-trained Generative Adversarial Network and used for unsupervised adaptation.