US-12619776-B2 - Anonymizing personally identifiable information in stored data

US12619776B2US 12619776 B2US12619776 B2US 12619776B2US-12619776-B2

Abstract

Obfuscating image data by receiving a plurality of video image frame data, detecting an object within a frame, generating a key associated with the object, matching the key to a key/value pair in a key store, generating a revised frame by replacing the object with the value, and providing the revised frame.

Inventors

ANJALI SHAH
NEIL DELIMA
SUMAN PATRA

Assignees

INTERNATIONAL BUSINESS MACHINES CORPORATION

Dates

Publication Date: 20260505
Application Date: 20210819

Claims (20)

1 . A computer implemented method for obfuscating image data, the method comprising: detecting, by one or more computer processors, an object within a frame of digital image data; generating, by the one or more computer processors, a key associated with a vector representation of the object, using a hash function; storing, by the one or more computer processors, the key in a key store matching, by the one or more computer processors, the key to a key/value pair in the key store; in response to determining that no matching the key/value pair exists in the key store, generating, by the one or more computer processors, a key/value pair comprising a pairing of the key and a value comprising a facial image generated using the detected object, in the key store; generating, by the one or more computer processors, a revised frame by replacing the object with a value from the matched key/value pair; and providing, by the one or more computer processors, the revised frame.
2 . The computer implemented method according to claim 1 , further comprising: generating, by the one or more computer processors, the value from the object; and storing, by the one or more computer processors, the value with the key in the key store.
3 . The computer implemented method according to claim 2 , wherein generating a value from the object comprises generating the value using a generative adversarial network.
4 . The computer implemented method according to claim 1 , further comprising generating, by the one or more computer processors, a first vector representation of the object from a first frame; and storing the first vector representation as the key in the key store.
5 . The computer implemented method according to claim 4 , wherein matching the key comprises: generating, by the one or more computer processors, a second vector representation of the object from a second frame; determining, by the one or more computer processors, a distance between the first vector representation and the second vector representation; and comparing, by the one or more computer processors, the distance to a threshold.
6 . The computer implemented method according to claim 1 , further comprising storing, by the one or more computer processors, the revised frame.
7 . The computer implemented method according to claim 1 , further comprising receiving, by the one or more computer processors, video image data; and parsing, by the one or more computer processors, the video image data into frames.
8 . A computer program product for obfuscating image data, the computer program product comprising one or more computer readable storage devices and collectively stored program instructions on the one or more computer readable storage devices, the stored program instructions comprising: program instructions to receive a plurality of video image frame data; program instructions to detect an object within a frame; program instructions to generate a key associated with a vector representation of the object, using a hash function; program instructions to store the key in a key store; program instructions to match the key to a key/value pair in the key store; program instructions to generate, in response to determining that no matching the key/value pair exists in the key store, a key/value pair comprising a pairing of the key and a value comprising a facial image generated using the detected object, in the key store; program instructions to generate a revised frame by replacing the object with a value from the matched key/value pair; and program instructions to provide the revised frame.
9 . The computer program product according to claim 8 , the stored program instructions further comprising: program instructions to generate the value from the object; and program instructions to store the value with the key in the key store.
10 . The computer program product according to claim 9 , wherein program instructions to generate a value from the object comprise program instructions to generate the value using a generative adversarial network.
11 . The computer program product according to claim 8 , the stored program instructions further comprising program instructions to generate a first vector representation of the object from a first frame; and program instructions to store the first vector representation as the key in the key store.
12 . The computer program product according to claim 11 , wherein the program instructions to match the key comprise: program instructions to generate a second vector representation of the object from a second frame; program instructions to determine a distance between the first vector representation and the second vector representation; and program instructions to compare the distance to a threshold.
13 . The computer program product according to claim 8 , the stored program instructions further comprising program instructions to store the revised frame.
14 . The computer program product according to claim 8 , the stored program instructions further comprising program instructions to receive video image data; and parsing the video image data into frames.
15 . A computer system for obfuscating image data, the computer system comprising: one or more computer processors; one or more computer readable storage devices; and stored program instructions on the one or more computer readable storage devices for execution by the one or more computer processors, the stored program instructions comprising: program instructions to receive a plurality of video image frame data; program instructions to detect an object within a frame; program instructions to generate a key associated with a vector representation of the object, using a hash function; program instructions to store the key in a key store; program instructions to match the key to a key/value pair in the key store; program instructions to generate, in response to determining that no matching the key/value pair exists in the key store, a key/value pair comprising a pairing of the key and a value comprising a facial image generated using the detected object, in the key store; program instructions to generate a revised frame by replacing the object with a value from the matched key/value pair; and program instructions to provide the revised frame.
16 . The computer system according to claim 15 , the stored program instructions further comprising: program instructions to generate the value from the object; and program instructions to store the value with the key in the key store.
17 . The computer system according to claim 16 , wherein program instructions to generate a value from the object comprise program instructions to generate the value using a generative adversarial network.
18 . The computer system according to claim 15 , the stored program instructions further comprising program instructions to generate a first vector representation of the object from a first frame; and program instructions to store the first vector representation as the key in the key store.
19 . The computer system according to claim 18 , wherein the program instructions to match the key comprise: program instructions to generate a second vector representation of the object from a second frame; program instructions to determine a distance between the first vector representation and the second vector representation; and program instructions to compare the distance to a threshold.
20 . The computer system according to claim 15 , the stored program instructions further comprising program instructions to store the revised frame.

Description

FIELD OF THE INVENTION The disclosure relates generally to anonymizing personally identifying information (PII) in stored data. The disclosure relates particularly to anonymizing object features captured in real-time image data. BACKGROUND Smart security system services have led to a proliferation of personal data being captured and stored as video streams. Biometric processing services that process face images and underlying face prints enable these security systems to identify individuals from such video streams. Generative adversarial networks (GAN) typically include two networks, a generator network and a discriminator network. The generator network creates outputs from input noise. The output is passed as an input to the discriminator network. The discriminator network then classifies that generated output as “real” data or as generated or “fake” data. During training the two networks have an adversarial relationship. The generator network seeks to maximize the value of the error associated with the discriminator network output, while the discriminator network seeks to minimize the value of the error. In a well-trained GAN, the discriminator network succeeds in correctly identifying generated output about 50% of the time and fails about 50% of the time. SUMMARY The following presents a summary to provide a basic understanding of one or more embodiments of the disclosure. This summary is not intended to identify key or critical elements or delineate any scope of the particular embodiments or any scope of the claims. Its sole purpose is to present concepts in a simplified form as a prelude to the more detailed description that is presented later. In one or more embodiments described herein, devices, systems, computer-implemented methods, apparatuses and/or computer program products enable obfuscation of personally identifiable information in video data. Aspects of the invention disclose methods, systems and computer readable media associated with obfuscating image data by receiving a plurality of video image frame data, detecting an object within a frame, generating a key associated with the object, matching the key to a key/value pair in a key store, generating a revised frame by replacing the object with the value, and providing the revised frame. BRIEF DESCRIPTION OF THE DRAWINGS Through the more detailed description of some embodiments of the present disclosure in the accompanying drawings, the above and other objects, features and advantages of the present disclosure will become more apparent, wherein the same reference generally refers to the same components in the embodiments of the present disclosure. FIG. 1 provides a schematic illustration of a computing environment, according to an embodiment of the invention. FIG. 2 provides a flowchart depicting an operational sequence, according to an embodiment of the invention. FIG. 3 depicts a cloud computing environment, according to an embodiment of the invention. FIG. 4 depicts abstraction model layers, according to an embodiment of the invention. DETAILED DESCRIPTION Some embodiments will be described in more detail with reference to the accompanying drawings, in which the embodiments of the present disclosure have been illustrated. However, the present disclosure can be implemented in various manners, and thus should not be construed to be limited to the embodiments disclosed herein. This description of the invention includes the example of the replacement of facial images in captured data as a non-limiting example. The scope of the disclosed embodiments may include the detection and replacement of faces, or other identifying objects in the captured data. Facial/object image data captured in video data streams may be processed for the purposes of facial recognition. Disclosed embodiments enable the storage of captured video data without the accompanying personally identifiable information of individual facial data. Aspects of the present invention relate generally to revising video data to obfuscate facial/object images rendering facial recognition impossible. Frames of video data are examined; faces and other identifying objects are detected, and generated data is used to replace the detected faces and/or objects on a frame-by-frame basis. Representations of individuals/objects etc. are captured; replacement images are generated and the pairs of original and replacement facial images are stored together. This enables the rapid replacement of those faces with generated images, without the computationally intensive need to continually generate new facial images for the same individual. In accordance with aspects of the invention there is a method for automatically detecting and replacing facial data with generated data to frustrate facial recognition efforts. The method reviews video image data frame by frame, each frame constituting the same time interval. Facial data and associated boundaries are detected. Vector representations of the facial data