IP Library › Granted Patent US 12,731,215
Granted Patent B2
US 12,731,215 · App. 18/588,816 · Granted Sep 8, 2026

Protecting personally identifiable information (PII) using monocular depth estimation

Inventors: Adrien David Gaidon (San Jose, CA); Vitor Campagnolo Guizilini (Santa Clara, CA)
Assignees: Toyota Research Institute, Inc.; Toyota Jidosha Kabushiki Kaisha
G06T5/20G06T7/50G06T2207/20081
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 12,731,215
App. No.
18/588,816
Granted
Sep 8, 2026
Kind
B2
Abstract

Systems, methods, and other embodiments described herein relate to protecting personally identifiable information (PII) with the use of a monocular depth estimation. In one embodiment, a method includes acquiring an original image depicting surrounding objects present in an environment. The method includes generating a depth map from the original image using a depth model that performs monocular depth estimation. The method includes obscuring at least a portion of the original image according to the depth map to provide an obscured image. The method includes providing the obscured image.

Claims (31)

1 . A masking system, comprising:

one or more processors;

a memory communicably coupled to the one or more processors and storing instructions that, when executed by the one or more processors, cause the one or more processors to:

acquire an original image depicting surrounding objects present in an environment;

generate a depth map from the original image using a depth model that performs monocular depth estimation;

obscure at least a portion of the original image according to the depth map to provide an obscured image including instructions to mask at least a portion of the surrounding objects by replacing the portion of the surrounding objects with corresponding portions from the depth map to remove personally identifiable information (PII); and

provide the obscured image.

2 . The masking system of claim 1 , wherein the depth model performs monocular depth estimation and is trained according to self-supervised structure-from-motion (SfM) training.

3 . The masking system of claim 1 , wherein the instructions include instructions to obscure the original image include instructions to replace the original image with the depth map and disposing of the original image to secure personally identifiable information (PII) in the original image.

4 . The masking system of claim 1 , wherein the instructions include instructions to obscure the original image include instructions to apply a semantic model to the original image that identifies semantic classes of the surrounding objects depicted in the original image, including at least people.

5 . The masking system of claim 1 , wherein the instructions to mask include instructions to place the portions of the depth map to obscure one or more of: faces and license plates.

6 . The masking system of claim 1 , wherein the instructions include instructions to provide the obscured image include instructions to provide the obscured image in place of the original image to secure personally identifiable information (PII).

7 . The masking system of claim 1 , wherein the masking system is integrated into an image processing pipeline within a vehicle to secure personally identifiable information (PII) in the original image during operation of the vehicle.

8 . A non-transitory computer-readable medium including instructions that, when executed by one or more processors, cause the one or more processors to:

acquire an original image depicting surrounding objects present in an environment;

generate a depth map from the original image using a depth model that performs monocular depth estimation;

obscure at least a portion of the original image according to the depth map to provide an obscured image including instructions to mask at least a portion of the surrounding objects by replacing the portion of the surrounding objects with corresponding portions from the depth map to remove personally identifiable information (PII); and

provide the obscured image.

9 . The non-transitory computer-readable medium of claim 8 , wherein the depth model performs monocular depth estimation and is trained according to self-supervised structure-from-motion (SfM) training.

10 . The non-transitory computer-readable medium of claim 8 , wherein the instructions include instructions to obscure the original image include instructions to replace the original image with the depth map and disposing of the original image to secure personally identifiable information (PII) in the original image.

11 . The non-transitory computer-readable medium of claim 8 , wherein the instructions include instructions to obscure the original image include instructions to apply a semantic model to the original image that identifies semantic classes of the surrounding objects depicted in the original image, including at least people.

12 . A method, comprising:

acquiring an original image depicting surrounding objects present in an environment;

generating a depth map from the original image using a depth model that performs monocular depth estimation;

obscuring at least a portion of the original image according to the depth map to provide an obscured image including masking at least a portion of the surrounding objects by replacing the portion of the surrounding objects with corresponding portions from the depth map to remove personally identifiable information (PII); and

providing the obscured image.

13 . The method of claim 12 , wherein the depth model performs monocular depth estimation and is trained according to self-supervised structure-from-motion (SfM) training.

14 . The method of claim 12 , wherein obscuring the original image includes replacing the original image with the depth map and disposing of the original image to secure personally identifiable information (PII) in the original image.

15 . The method of claim 12 , wherein obscuring the original image includes applying a semantic model to the original image that identifies semantic classes of the surrounding objects depicted in the original image, including at least people.

16 . The method of claim 12 , wherein masking includes placing the portions of the depth map to obscure one or more of: faces and license plates.

17 . The method of claim 12 , wherein providing the obscured image includes providing the obscured image in place of the original image to secure personally identifiable information (PII).

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Mar 12, 2024
From: GAIDON, ADRIEN DAVID; CAMPAGNOLO GUIZILINI, VITOR
To: TOYOTA RESEARCH INSTITUTE, INC.; TOYOTA JIDOSHA KABUSHIKI KAISHA
Reel/Frame 066732/0362 →
Continuity (1)
Related Publication 20250272788A1 · Aug 28, 2025
References Cited (10)
US 11157646B2 · Grau · 2021 [cited by applicant]
US 20100007912A1 · Saito · 2010 [cited by applicant]
US 20170104893A1 · Hardy et al. · 2017 [cited by applicant]
US 20190377901A1 · Balzer et al. · 2019 [cited by applicant]
US 20200285771A1 · Dey et al. · 2020 [cited by applicant]
US 20210004974A1 · Guizilini · 2021 [cited by examiner]
US 20210312079A1 · Nishida et al. · 2021 [cited by applicant]
US 20230281336A1 · Pottier et al. · 2023 [cited by applicant]
US 20240394389A1 · Wen · 2024 [cited by examiner]
US 20250131538A1 · Attebo · 2025 [cited by examiner]