IP Library › Granted Patent US 12,738,018
Granted Patent B2
US 12,738,018 · App. 18/186,061 · Granted Sep 15, 2026

Safety monitor for image misclassification

Inventors: Tung Chuen Kwong (Markham, CA); Benjamin Koon Pan Chan (Markham, CA); David Porpino Sobreira Marques (Kleinburg, CA); Clarence Ip (Markham, CA); Hung Wilson Yu (Markham, CA)
Assignee: ATI Technologies ULC
G06V10/454G06F17/18G06F18/24G06N5/04G06V10/764G06V10/776G06V10/82G06V10/98G06V20/10
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,738,018
App. No.
18/186,061
Granted
Sep 15, 2026
Kind
B2
Abstract

Systems, apparatuses, and methods for implementing a safety monitor framework for a safety-critical inference application are disclosed. A system includes a safety-critical inference application, a safety monitor, and an inference accelerator engine. The safety monitor receives an input image, test data, and a neural network specification from the safety-critical inference application. The safety monitor generates a modified image by adding additional objects outside of the input image. The safety monitor provides the modified image and neural network specification to the inference accelerator engine which processes the modified image and provides outputs to the safety monitor. The safety monitor determines the likelihood of erroneous processing of the original input image by comparing the outputs for the additional objects with a known good result. The safety monitor complements the overall fault coverage of the inference accelerator engine and covers faults only observable at the network level.

Claims (47)

1 . A system comprising:

a processor comprising circuitry configured to:

track a frequency of detection of one or more objects in previously received image data;

add at least one object of the one or more objects to current image data received subsequent to the previously received image data by inserting pixel data representing the at least one object into the current image data to generate modified image data, responsive to a frequency of detection of the at least one object in the previously received image data exceeding a first threshold; and

generate an indication corresponding to a calculated likelihood of misclassification of the current image data, based at least in part on processing the modified image data; and

perform an action in response to the indication.

2 . The system as claimed in claim 1 , wherein the processor is configured to add the at least one object outside a boundary of an image corresponding to the current image data, to generate the modified image data.

3 . The system as claimed in claim 1 , wherein the processor is configured to generate the indication based at least in part on an output of a machine learning model.

4 . The system as claimed in claim 1 , wherein the processor is further configured to:

cause the modified image data to be processed by transmitting the modified image data to an inference accelerator; and

receive, from the inference accelerator, at least one output indicative of processing of the modified image data by the inference accelerator.

5 . The system as claimed in claim 4 , wherein the current image data is deemed misclassified, responsive to the at least one output not being consistent with a given test vector data corresponding to the current image data.

6 . The system as claimed in claim 1 , wherein the at least one object, represented by the pixel data, is selected based at least in part on a given test vector data corresponding to the previously received image data.

7 . The system as claimed in claim 1 , wherein the action comprises causing a safety-critical application to perform at least one of:

terminating operation;

re processing a same image;

rebooting a system;

generating a warning signal;

reducing a speed of a vehicle; or

changing an operating mode.

8 . A method comprising:

tracking, by a processor comprising circuitry, a frequency of detection of one or more objects in previously received image data; and

adding, by the processor, at least one object of the one or more objects to current image data received subsequent to the previously received image data by inserting pixel data representing the at least one object into the current image data to generate modified image data, responsive to a frequency of detection of the at least one object in the previously received image data exceeding a first threshold;

generating an indication corresponding to a calculated likelihood of misclassification of the current image data, based at least in part on processing the modified image data; and

performing an action in response to the indication.

9 . The method as claimed in claim 8 , further comprising adding, by the processor, the at least one object outside a boundary of an image corresponding to the current image data, to generate the modified image data.

10 . The method as claimed in claim 8 , further comprising generating the indication based at least in part on an output of a machine learning model.

11 . The method as claimed in claim 8 , further comprising:

causing, by the processor, the modified image data to be transmitted to an inference accelerator; and

receiving, by the processor from the inference accelerator, at least one output indicative of processing of the modified image data by the inference accelerator.

12 . The method as claimed in claim 11 , wherein the current image data is deemed misclassified, responsive to the at least one output not being consistent with a given test vector data corresponding to the current image data.

13 . The method as claimed in claim 8 , wherein the at least one object, represented by the pixel data, is selected based at least in part on a given test vector data corresponding to the previously received image data.

14 . The method as claimed in claim 11 , wherein the at least one object is identified as an object having a calculated probability of being correctly identified by the inference accelerator in the previously received image data, exceeding a second threshold.

15 . A system comprising:

an inference accelerator comprising circuitry; and

a processor comprising circuitry configured to:

track a frequency of detection of one or more objects in previously received image data;

add at least one object of the one or more objects to current image data received subsequent to the previously received image data by inserting pixel data representing the at least one object into the current image data to generate modified image data, responsive to a frequency of detection of the at least one object in the previously received image data exceeding a first threshold;

generate an indication corresponding to a calculated likelihood of misclassification of the current image data, based at least in part on processing of the modified image data by the inference accelerator; and

perform an action in response to the indication.

16 . The system as claimed in claim 15 , wherein the processor is configured to add the at least one object outside a boundary of an image corresponding to the current image data, to generate modified image data.

17 . The system as claimed in claim 15 , wherein the inference accelerator is configured to execute a machine learning model.

18 . The system as claimed in claim 15 , where in the processor is further configured to:

transmit the modified image data to the inference accelerator; and

receive, from the inference accelerator, at least one output indicative of processing of the modified image data by the inference accelerator.

19 . The system as claimed in claim 15 , wherein the at least one object is selected based at least in part on a given test vector data corresponding to the previously received image data.

20 . The system as claimed in claim 15 , wherein the at least one object is identified as an object having a calculated probability of being correctly identified by the inference accelerator in previously processed image data.

Continuity (2)
Continuation 16424162 · May 28, 2019
Related Publication 20230230367A1 · Jul 20, 2023
References Cited (116)
US 5020112A · Chou · 1991 [cited by applicant]
US 5657420A · Jacobs et al. · 1997 [cited by applicant]
US 6067287A · Chung-Ju et al. · 2000 [cited by applicant]
US 6215915B1 · Reyzin · 2001 [cited by applicant]
US 6295374B1 · Robinson et al. · 2001 [cited by applicant]
US 6678404B1 · Lee et al. · 2004 [cited by applicant]
US 8131660B2 · Davis et al. · 2012 [cited by applicant]
US 8345979B2 · Davis · 2013 [cited by applicant]
US 8369595B1 · Derakhshani et al. · 2013 [cited by applicant]
US 9319137B2 · Zhuge et al. · 2016 [cited by applicant]
US 9501811B2 · Michel · 2016 [cited by applicant]
US 10078794B2 · Pierce et al. · 2018 [cited by applicant]
US 10339622B1 · Tang et al. · 2019 [cited by applicant]
US 10438091B2 · Li · 2019 [cited by applicant]
US 10726723B1 · Madden · 2020 [cited by examiner]
US 10762392B2 · Zhang et al. · 2020 [cited by applicant]
US 11593644B2 · Miniskar · 2023 [cited by examiner]
US 11610142B2 · Kwong et al. · 2023 [cited by applicant]
US 20030086616A1 · Oh et al. · 2003 [cited by applicant]
US 20040263911A1 · Rodriguez et al. · 2004 [cited by applicant]
US 20050104976A1 · Currans · 2005 [cited by applicant]
US 20060269119A1 · Goldberg et al. · 2006 [cited by applicant]
US 20070036420A1 · Enachescu et al. · 2007 [cited by applicant]
US 20070258640A1 · Abe et al. · 2007 [cited by applicant]
US 20080075378A1 · Kim et al. · 2008 [cited by applicant]
US 20080211915A1 · McCubbrey · 2008 [cited by applicant]
US 20150178246A1 · Herrero Abellanas et al. · 2015 [cited by applicant]
US 20150178889A1 · Michel · 2015 [cited by applicant]
US 20150206337A1 · Roimela · 2015 [cited by applicant]
US 20150278601A1 · Nagamine · 2015 [cited by examiner]
US 20150304648A1 · Gulati et al. · 2015 [cited by applicant]
US 20150350638A1 · Beck · 2015 [cited by applicant]
US 20150358755A1 · Luo · 2015 [cited by applicant]
US 20160062294A1 · Murashima · 2016 [cited by applicant]
US 20160092736A1 · Mai · 2016 [cited by examiner]
US 20160104070A1 · Eslami · 2016 [cited by examiner]
US 20160179434A1 · Herrero Abellanas et al. · 2016 [cited by applicant]
US 20160239682A1 · Kapadia · 2016 [cited by examiner]
US 20160245178A1 · Bhabhrawala · 2016 [cited by applicant]
US 20170177458A1 · Viggers et al. · 2017 [cited by applicant]
US 20170206415A1 · Redden · 2017 [cited by applicant]
US 20170316312A1 · Goyal · 2017 [cited by applicant]
US 20180018864A1 · Baker · 2018 [cited by applicant]
US 20180032859A1 · Park et al. · 2018 [cited by applicant]
US 20180068429A1 · Tan et al. · 2018 [cited by applicant]
US 20180173698A1 · Dubey et al. · 2018 [cited by applicant]
US 20180189641A1 · Boesch et al. · 2018 [cited by applicant]
US 20180218303A1 · Cole et al. · 2018 [cited by applicant]
US 20180218502A1 · Golden · 2018 [cited by examiner]
US 20180223641A1 · Ciezobka · 2018 [cited by examiner]
US 20180247412A1 · Gooding et al. · 2018 [cited by applicant]
US 20180330166A1 · Redden et al. · 2018 [cited by applicant]
US 20180373941A1 · Kwant et al. · 2018 [cited by applicant]
US 20190028752A1 · Zhang et al. · 2019 [cited by applicant]
US 20190050685A1 · Kaminski et al. · 2019 [cited by applicant]
US 20190080205A1 · Kaufhold · 2019 [cited by examiner]
US 20190106117A1 · Goldberg · 2019 [cited by applicant]
US 20190108384A1 · Wang et al. · 2019 [cited by applicant]
US 20190122384A1 · Ertle · 2019 [cited by examiner]
US 20190147220A1 · Mccormac · 2019 [cited by examiner]
US 20190147332A1 · Agudu et al. · 2019 [cited by applicant]
US 20190164265A1 · Liao et al. · 2019 [cited by applicant]
US 20190183456A1 · Savadjiev · 2019 [cited by examiner]
US 20190220381A1 · Pogorelik · 2019 [cited by examiner]
US 20190258251A1 · Ditty et al. · 2019 [cited by applicant]
US 20190318099A1 · Carvalho · 2019 [cited by examiner]
US 20190325305A1 · Zhang et al. · 2019 [cited by applicant]
US 20190332897A1 · Chen et al. · 2019 [cited by applicant]
US 20190332926A1 · Lie · 2019 [cited by examiner]
US 20200005422A1 · Subramanian et al. · 2020 [cited by applicant]
US 20200037987A1 · Silberman et al. · 2020 [cited by applicant]
US 20200134432A1 · Lagudu et al. · 2020 [cited by applicant]
US 20200302285A1 · Wang et al. · 2020 [cited by applicant]
US 20200327378A1 · Smith et al. · 2020 [cited by applicant]
EP 3098762A1 · 2016 [cited by applicant]
EP 3467662A1 · 2019 [cited by applicant]
EP 3475875A1 · 2019 [cited by applicant]
JP 2007096613A · 2007 [cited by applicant]
JP 2013196265A · 2013 [cited by examiner]
JP 2015191334A · 2015 [cited by applicant]
JP 2016085721A · 2016 [cited by applicant]
JP 2018005339 · 2018 [cited by applicant]
JP 2020061066A · 2020 [cited by applicant]
WO 2014155639A · 2014 [cited by applicant]
WO 2017003887A1 · 2017 [cited by applicant]
WO 2017075939A1 · 2017 [cited by applicant]
WO 2019094843A1 · 2019 [cited by applicant]
International Search Report and Written Opinion in International Application No. PCT/IB2020/054448, mailed Aug. 14, 2020, 9 pages. [cited by applicant]
Non-Final Office Action in U.S. Appl. No. 15/657,613, mailed Oct. 5, 2018, 12 pages. [cited by applicant]
Final Office Action in U.S. Appl. No. 15/657,613, mailed Mar. 8, 2019, 10 pages. [cited by applicant]
International Search Report and Written Opinion in International Application No. PCT/US2018/052358, mailed Feb. 18, 2019, 13 pages. [cited by applicant]
Cecconi et al., “Optimal Tiling Strategy for Memory Bandwidth Reduction for CNNs”, International Conference on Advanced Concepts for Intelligent Vision Systems, Sep. 18, 2017, pp. 89-100. [cited by applicant]
Fan et al., “F-C3D: FPGA-based 3-Dimensional Convolutional Neural Network”, 27th International Conference on Field Programmable Logic and Applications (FPL), Sep. 4, 2017, 4 pages. [cited by applicant]
Rahman et al., “Efficient FPGA Acceleration of Convolutional Neural Networks Using Logical-3D Compute Array”, Proceedings of the 2016 Conference on Design, Automation & Test in Europe, Mar. 14, 2016, pp. 1393-1398. [cited by applicant]
Notice of Allowance in U.S. Appl. No. 16/234,956, mailed May 5, 2020, 10 pages. [cited by applicant]
“BlackBerry QNX Platform for Instrument Clusters”, BlackBerry QNX, 2018, 4 pages, https://blackberry.qnx.com/content/dam/qnx/products/instrument-clusters/instrument-clusters-product-brief.pdf. [Retrieved Jun. 17, 2019]. [cited by applicant]
Zander, Justyna, “Functional Safety for Autonomous Driving”, IS&T Symposium on Electronic Imaging 2017; Autonomous Vehicles and Machines 2017, Jan. 29-Feb. 2, 2017, 29 pages, Nvidia, https://www.imaging.org/Site/PDFS/Co… [cited by applicant]
Non-Final Office Action in U.S. Appl. No. 16/427,941, mailed Apr. 29, 2021, 14 pages. [cited by applicant]
“Appendix 1 to the Oct. 2019 Update: Subject Matter Eligibility, Life Sciences & Data Processing Examples”, Oct. 2019, 41 pages. [cited by applicant]
International Search Report and Written Opinion in International Application No. PCT/IB2020/054645, mailed Aug. 21, 2020, 8 pages. [cited by applicant]
Notice of Allowance in U.S. Appl. No. 16/427,941, mailed Aug. 31, 2021, 9 pages. [cited by applicant]
Non-Final Office Action in U.S. Appl. No. 17/548,393, mailed Oct. 4, 2022, 14 pages. [cited by applicant]
Tian, Yuchi at Virginia Edu et al., “Deep Test automated testing of deep-neural-network-driven autonomous cars”, Software Engineering, ACM, 2 Penn Plaza, Suite 701 New York NY 10121-0701 USA, May 27, 2018, pp. 303-314. [cited by applicant]
Dwibedi, Debidatta et al., “Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detection”, 2017 IEEE International Conference on Computer Vision, Oct. 22, 2017, pp. 1310-1319. [cited by applicant]
Xiaofei, Xie, et al.; “DeepHunter: Hunting Deep Neural Network Defects via Coverage-Guided Fuzzing”, Nov. 16, 2018; Retrieved from the Internet: URL: https://arxiv.org/pdf/1809.01266.pdf. [cited by applicant]
European Search Report European Application No. 20813760.4, mailed May 22, 2023, 12 pages. [cited by applicant]
Supplementary European Search Report in EP Application No. 20812662.3, mailed Oct. 30, 2023, 10 pages. [cited by applicant]
Final Office Action for U.S. Appl. No. 17/548,393, dated May 4, 2023, 7 pages. [cited by applicant]
Office Action for JP Patent Application No. 2021-570761, dated Sep. 5, 2023, 6 pages. [cited by applicant]
Office Action for KR Patent Application No. 10-2021-7042278, dated Sep. 14, 2023, 15 pages. [cited by applicant]
Notice of Allowance for U.S. Appl. No. 17/548,393, dated Sep. 11, 2023, 7 pages. [cited by applicant]
Suman Jana et al., DeepTest: Automated Testing of Deep-Neural-Network-drivenAutonomous Cars, ICSE'18, May 27-Jun. 3, 2018, Gothenburg, Sweden. [cited by applicant]
Office Action for JP Patent Application No. 2021-570317, dated Mar. 5, 2024, 8 pages. [cited by applicant]
Office Action for JP Patent Application No. 2024-013261, dated Mar. 5, 2024, Dec. 10, 2024. [cited by applicant]
Tian Yuchi et al. , “DeepTest: Automated Testing of Deep-Neural-Network-Driven Autonomous Cars” , 2018 IEEE/ACM 40th International Conference on Software Engineering (ICSE)[online] , IEEE , 2018 , pp. 303-314 , [Search … [cited by applicant]
Shimizu Naoshige, GM Driverless car defeat google strongest contender 2019, mass production, Nikkei Automotive No. 85, Nikkei BP Ltd., Mar. 11, 2018, pp. 50-56. [cited by applicant]