Automated FaceBlurring feature
Anonymization techniques are crucial for protecting individuals' privacy involved in projects, such as workers or clients, and complying with data protection regulations. Here are five key reasons why they are important:
Privacy Protection: Anonymization ensures that personally identifiable information (PII) such as a person’s face, body or other identifying features are removed or obfuscated from datasets. This helps prevent unauthorized access to sensitive information and minimizes the risk of privacy breaches.
Legal Compliance: Many data protection regulations, such as the General Data Protection Regulation (GDPR) in the European Union and the Health Insurance Portability and Accountability Act (HIPAA) in the United States, require organizations to anonymize personal data to safeguard individuals' privacy rights. Failure to comply with these regulations can result in significant fines and legal penalties.
Data Sharing: Anonymization facilitates the sharing of data for research, analysis, and other purposes without compromising individuals' privacy. By anonymizing sensitive information, organizations can share datasets with researchers, government agencies, and other stakeholders while minimizing the risk of re-identification.
Ethical Considerations: Respecting individuals' privacy is not only a legal requirement but also an ethical obligation. Anonymization helps organizations uphold ethical standards by protecting individuals' rights to privacy and anonymity.
Risk Mitigation: Anonymization reduces the risk of data breaches and unauthorized access by limiting the amount of personally identifiable information available in datasets. This mitigates the potential harm that could result from data leaks, identity theft, or other privacy violations.
This is why feature such as “FaceBlurring” play a crucial role in safeguarding individuals' privacy, ensuring legal compliance, facilitating data sharing, upholding ethical standards, and mitigating risks associated with data breaches and unauthorized access.
How to set up automated FaceBlurring?
Automatic FaceBlurring is a special feature that needs to be activated for your workspace upon request - please contact your account manager to get access or reach out to support@holobuilder.com for more information.
How does the automated FaceBlurring work?
Face Blurring is a feature intended to safeguard the identities of individuals who may appear in your 360° video or photo captures. When activated, every face detected will be masked (pixelated by default) after uploading and publishing to your project.
Boundaries of the System
The masking step is done in post-processing and alters the captures irreversibly. This privacy function operates through a machine learning algorithm, and the accuracy of its results depends on various factors. Although this procedure employs advanced machine learning algorithms to accurately recognize generic individual, it does not pinpoint specific faces or individuals, and it cannot completely eliminate the rate of false-negatives or false-positives.
Faces Remain Unblurred - "False Negatives"
When a face remains unblurred, it typically indicates that the facial recognition system is unable to discern a face due to factors in the image, like distortion, pixelation, surgical masks, or other obstructions. This outcome is a common feature of the machine learning algorithms we utilize for person detection. We refine our models to minimize both false negatives and false positives, aiming to provide exceptional service to customers. However, in this process, certain resemblances, while statistically unrecognizable, might still appear unblurred and identifiable to viewers familiar with the subject or surroundings.
Unintended Parts of the Images are Blurred – “False Positives”
If parts of the image are blurred where faces do not exist, that usually means the facial recognition incorrectly identified a face where there was not one. This is a result of the machine learning algorithms we use to detect a person. We tune our models to reduce both false positives and false negatives to deliver a superior service to customers, but in doing so, some objects or aspects of images captured may be blurred incorrectly. We value and take data protection policies very seriously, and our policy is to be stricter with what we are blurring vs. running the risk of not blurring a person.
How to get Involved?
At FARO, our commitment to innovation drives us to continuously enhance our AI product, integrating cutting-edge technologies, refining algorithms, and responding to user feedback to deliver ever-improving performance, functionality, and value. We embrace the support of our customers and community, and here is how you can support us create better products and services for you.
Mask UnBlurred Area Manually
In the WebEditor, you have the option to mask any area you want to hide manually. Please find more information about how to here → How to use manual blurring on your 360 scenes.
Report Issues to Us
Reporting issues with Face Blurring helps us improve the algorithms. Please contact our support team at support@holobuilder.com or support-eu@holobuilder.com. We only need a few details to investigate the issue effectively:
What type of error occurred (false-negative/false-positive)?
Where does the error occur?
Please provide the project name and link to it (from the browser's address bar)
Please add a link to the unblurred image
Provide example Image to train our System
By default, we only process 360° video or photo captures to mask them; we do not analyze them any further in any case. However, with explicit consent and approval from the owner of the data, we can train our system to be more effective at identifying PII, e.g., faces and bodies, in certain situations which might not be covered yet. If you wish to provide such information, please contact us at support@holobuilder.com or support-eu@holobuilder.com. We are grateful for your support!
Best-practices to Increase Privacy on your Site
Inform your Team Members & Visitors
As a customer, it is your responsibility to seek consent from individuals whose data may be captured (such as team members or visitors to your site/projects). For instance, you can inform them through notices that 360° photo documentation will occur, and measures will be implemented to de-identify facial recognition.
Scheduling Captures During Non-Work Hours
Additionally, you can take proactive steps during image capture to minimize or avoid collecting data from individuals, such as scheduling captures during non-work hours when there are fewer people present at the location.