Discover the new assisted video redaction feature in the latest Amped Replay release, which uses machine learning to quickly find and hide people, heads, vehicles, and license plates in your video evidence, while keeping you in the driving seat. Learn about the workflow, reporting safeguards, tips for annotation, and other new format support and performance improvements.
Hello everyone and welcome back to a groundbreaking release of Amped Replay, the forensic video player of choice for law enforcement officers and investigators!
Why dare I say groundbreaking, you may ask? Because we have developed assisted redaction, an embedded machine learning-based system, to assist with the finding and redacting of vehicles and persons in video. And we launch the new feature with this release!
As well as the new redaction features, we have also introduced support for a number of new video formats and variants. Read on to find out more!
See the New Features in Action
Assisted Video Redaction in Amped Replay
How Do I Redact Video Evidence with Assisted Redaction in Amped Replay?
If you know us a little, you’ll also know how cautious we are and always have been when it comes to using AI in our products. We understand the risks associated with this technology, especially when it comes to image/video enhancement. However, when it is used to assist in redacting sensitive content in video material, such as persons or vehicles, there is no doubt that it can be of great benefit, as well as being forensically safe.
When combined with a careful human review, machine learning methods can significantly reduce your workload. Amped Replay is the perfect product for such a model, because one of its main functions is to assist officers and investigators in redacting sensitive information in videos. Well, that’s where assisted redaction comes into play. Using a machine learning system, you can now safely and easily find and redact persons, human heads, vehicles and license plates in your video evidence! We all know how tedious it can be to blur out or pixelate identifiable individuals and vehicles in a body worn video, when both the camera and the subjects/objects are moving like crazy in all directions.
Step-by-step: How to Run Assisted Redaction in Amped Replay
The burden is now on the machine learning system, which can be more accurate and motivated than any human around! Using it could not be any easier, either. First, load your video and if necessary, select an appropriate range to which the assisted redactions should be applied to. Working with a range to start with means that the process of finding subjects and vehicles later will be that much quicker.
Remember that if your footage is too dark or lacks good contrast, you can still use the Enhance tab to improve the pixel values with the light filters. And similarly, you can perform any enhancement of choice before or after having done the intended redactions. Click on the Redact tab and then click on the Assisted Redaction icon.
You can search for persons (full body), heads (faces, backs of heads and profiles), vehicles (of all kinds) and license plates. Tick the options as required and if you wish, you can also add textual labels to each redaction by using the “Add Labels” tick box. Now, click on the “Run Assisted Redaction” button. As soon as this is completed, a new automatic Sort Assisted Video Redaction panel will appear on the left of the interface, displaying the results of the search.
Review, Customize, and Merge Your Video Redactions
The window displays all the persons, heads, vehicles and license plates found, labelled individually (i.e. Vehicle-1, Person-2 etc.). By clicking on each entry in the list, you will get a preview on viewer. In this panel, you can remove items you do not want to redact by clicking on the corresponding tick boxes and then clicking on the “Remove” button. Finally, click on the “Apply All” button to perform the redactions as indicated.
In the example below, you can see that the subject removed has not been redacted, whilst the rest have been pixelated. You’ll also now notice an activity plot in the timeline. This gives a visual account of how many objects are detected and redacted by the system (highlighted in yellow below).
If you wish, you can also customize all the Hide objects by double-clicking on any redaction and by changing the Strength and other parameters as per your preference. All you have to do then is right-click and select the option “Apply properties to all Hide”. You can also customize the text labels (font, size, color etc.) in the same manner. In the sample below, I have changed the hide ellipse to be circumscribed and applied to all redactions, so that no detected face will reveal any identifiable features.
You can also merge two or more items by ticking them and then clicking on the “Merge” button. This option will be useful when the system detects the same subject or vehicle as being different. In the example below, subjects have been merged when they were visually identified as the same.
Once you have merged all the relevant subjects/objects, you will end up with a streamlined list containing the individual subjects or vehicles. Then, you can apply some final customizations, such as the textual labels. You can edit these simply by clicking on them in the panel and typing the chosen label with the keyboard. You can also change the class type in the same manner.
Forensic Reporting and Human Review
Once you have finished, you can go to the Export tab and finalize your evidence. And now an important point: we named this feature Assisted Redaction because it’s meant to assist you, and not replace you. We strongly recommend reviewing the video evidence and checking that the redactions have been applied correctly. If the redactor missed any frames you would like to hide, you can, of course, click on individual redactions and refine using the conventional Replay key framing or tracking features. The usage of the new machine learning system is thoroughly documented in the report generation, so that it is properly audited and disclosed alongside other processes used by the program.
Not ready for this new feature? Or maybe your organization does not yet allow any use of AI for investigative purposes? No problem. With this release, we also provide an installer without this feature so you can still enjoy the other updates and remain compliant with your requirements.
Tip: How to Use the New AI Model to Annotate!
At the moment, the new AI model is only applicable to the Redact tab. However, there is a nice little workaround that you can use to annotate subjects and objects using the new feature.
All you have to do is track and redact your target using the procedure explained earlier. Then:
- right-click on the redaction and select the option “Copy Tracking Data”
- delete the redaction, click on the Annotate tab, draw the selected annotation, right-click on it and select “Paste Tracking Data”
The target will now be tracked exactly as it was when you applied the temporary assisted redaction. Pretty cool, right?
Formats
In this update, we have added support for a number of new formats and variants! More specifically:
- 3 new supported formats
- 1 new variations of already-supported formats
Remember: if you find a video file we don’t fully support, there is a good chance we have never seen it before! Hop over to the Amped Support Portal, upload it, and we will get on it. While doing that, make sure to provide as much information as possible about the DVR’s brand and model. Also, provide the proprietary player, if available. Submitting new formats not only helps you out but also the entire Amped community. The best part? This is all included in an active Software and Maintenance Support plan. Learn here why it is important to keep your SMS up to date.
Other Improvements
We continue to improve performance and background functionality in every new release of Amped Replay. This is what we have also updated:
- The program manual has been updated.
- When installing or re-installing the program, the installer will check for the correct location and functioning of all embedded executables, warning the user if any issues are found.
- The internal video libraries have been updated to the newer version.
- The export functions now display tooltips when hovering over the buttons with the mouse.
Bugs
In this release, we have fixed a number of minor issues reported by our trusted users.
- Fixed a graphical interface issue occurring when switching Windows user accounts.
- Fixed a bug that caused the program to crash when processing too many redact or annotate objects.
- Fixed a bug that caused the Annotate tab to be displayed when loading audio-only files.
The Final Word
Replay overcomes the challenges of a limited toolset often encountered by police officers and detectives worldwide. It protects them with the forensic rigor required in today’s legal environment and speeds up investigations.
With the new Assisted Redaction feature, officers and investigators can quickly locate and redact people, vehicles, and license plates in video evidence while maintaining full control of the process.
Forensic video analysts and technicians have benefited from Amped FIVE for many years. It is now possible to perform the most basic enhancement tasks and annotate images and videos with Amped Replay, too. If you are lucky enough to have Replay on your desktop, then update today or email us.
Don’t Delay – Update Today
If you have an active support plan, you can update straight away by going into the menu About > Check for Updates within Amped Replay. If you need to renew your SMS plan, please contact us or one of our authorized distributors. And remember that you can always manage your license and requests from the Amped Support Portal.
FAQs – Assisted Redaction in Amped Replay
Assisted Redaction is a machine learning-based system embedded in Amped Replay that automatically detects people, heads, vehicles, and license plates so you can quickly redact them in video evidence while keeping full human control.
Assisted Redaction can find and track full-body persons, human heads, vehicles, and license plates across your video evidence.
Yes. The use of the Assisted Redaction feature is fully documented in Amped Replay’s report. It is, however, a tool designed to assist the user; we strongly recommend that the user double-checks the output to correct any possible mistakes.
No. It is designed to assist you, not replace you. You should always review the video, refine any redactions if needed, and confirm that sensitive content has been properly obscured.
Yes. If your organization does not yet allow AI for evidential purposes, an installer without the Assisted Redaction feature is available so you can still benefit from the other updates.