How does Secure Redact work?
Secure Redact is powered by a type of AI called deep learning, the system's algorithms have been specially trained to automatically detect personal data in security video footage. All you need to do to get started is upload a video and the Secure Redact system will automatically find all the faces (number plates are currently being added) in each frame of your video (or image), you can then quickly check the results and either blur all detected content or select which people to anonymise. The redacted video can then be downloaded and used as required.
Are there any system requirements to use Secure Redact?
Secure Redact is a cloud processing solution, the application runs in a local web browser. Secure Redact supports web browsers: Chrome & Firefox
Which video formats does Secure Redact support?
All major video file formats including 'mp4', 'mov', 'avi', 'webm', 'wmv', 'mpeg'. If you have a file format that is not currently supported please get in contact, firstname.lastname@example.org, and we will look into adding it.
Can I upload individual images?
Yes, Secure Redact can run on images as well as video files.
How long does it take to process a video?
There are four stages to processing: 1. Initial video upload, this depends on the upload bandwidth of your connection 2. Conversion of your video into a format suitable for processing 3. Automatic detection of content in video 4. Redacting video footage As a rule of thumb overall processing should take around 2x longer than the video, this will vary depending on the frame rate of your video.
What if performance is not as expected?
CCTV and other security footage tends to include noisy scenes with high levels of occlusion and movement. Video is captured on a wide variety of lenses, camera angles, camera mounting positions and with varying resolution/quality. These variances across footage can have an impact on Secure Redact's ability to detect people and number plates if a more unique video angle or context is uploaded. If face and number-plate detection performance is not as expected for your video please get in contact, email@example.com, and we will endeavour to update our algorithms to take into account your specific style of content.
Will the system pick up 100% of all faces and number plates?
Detecting all faces and number plates in all frames in a video is a difficult challenge even for a machine. An hour of video footage from a busy scene can contain around 2 million individual faces (if you add all faces that appear in each frame together); even at 99.9% accuracy an algorithm would still miss 2k faces – missed faces may occur from acute changes in head position, lighting, movement and occlusion. Very small faces may also be missed if they are unrecognisable due to low resolution. The application has a suite of editing tools so you can quickly review overlaid detections on each video. If required you can then quickly add additional detection boxes or individual tracks. If you do pick-up any missed detections please let us know as we are continually improving service performance for these cases: firstname.lastname@example.org.
Where can I find the user guide?
The user guide is accessed via the menu in the application.
Can Secure Redact only be used as a SaaS solution?
If you have needs for alternative implementations of video redaction please contact email@example.com with your requirements.
How does the Secure Redact pricing work?
Secure Redact can be accessed through a monthly subscription or by purchasing video credits. Monthly subscriptions:- there are a range of monthly subscription tiers based on the amount of video that needs processing each month. Subscriptions can be cancelled or upgraded at any time, changes will take effect the next month. Credits:- video credits can also be purchased for a specific amount of video, credits last 6 months from the date of purchase. Video usage is metered on upload. Single images are metered as the equivalent of 1 second of video.