Need To Blur / Mask People On Exported Video - How?

Hi all,

Is anyone aware of a way to mask objects or persons in video post export? Our use case is blurring or masking associates or customers faces that maybe in a store during an incident. We would be ok with initially blurring in either Windows Media Player or VLC as we understand the potential limitations to blurring in the original format. Thanks in advance...


Rodney, typically video editing software allows adding such overlays to blur/mask specific objects. The question, then, becomes which one is the easiest to use for you?

Also, one other consideration would be if that impacts its admission into court. This probably can be handled simply by keeping two copies - an original unmasked / unblurred one and the second produced one.

Thanks for the response John. Most of the masking I have seen is engineered to occur prior to pulling the video. This is effective if you are trying to blur out a specific object etc but is typically not helpful when trying to blur a lobby full of customers. My hope is to find something that will allow us to pull (export) and then blur specific areas (faces) after the fact.

Regarding the watermarking issues, we do plan on using two different copies. One would be the intial WMP or VLC blurred version that Law Enforcement would use to get a description out and the other, watermarked version would be used in court.

Thanks again

YouTube has an online video editor that's actually quite functional, and it includes the ability to blur faces automatically (see article here). If the cops are going to release clips via YouTube, might be an idea to work with them that way, either posting and blurring yourself and giving them the link, or just giving them the clip and letting them do it.

You can also upload the video privately, do the blurring, then re-download it and pass it on.

I was digging that YouTube suggestion Matt, but looks like that blurs ALL faces not just the ones you choose.

Matt/Ross - thanks for the feedback. I tried the YouTube option but agree with Matt that it is an all or nothing solution. In reviewing further, I did found a potential solution via Adobe Premier Pro but, as you can see by the instructional video, it is quite an extensive process to blur a face in a video clip.

I am a bit surprised that there is not an easier solution as this seems like a logical need for retailers sharing video with LE.

Thanks again for the responses.

You can blur / mask objects with most basic video editing software (e.g., Camtasia). I bet higher end ones can track objects and keep the blur centered. Basic ones just let you put a static mask over the area and if the person moves substantially, you need to manually move the blur/mask. However, many/most cases / evidence that are typically short with minimal activity to blur/mask.

I have just come accross a bid that requires exactly this function - they want to blur the video of innocent bystanders post-pocessing.

Does anyone know of a VMS that has this ability built-in (where you do not need to use third-party video editing software)?

Perhaps there is a video analytics add-on that could do this to an exported video clip?

A, I don't know of a manufacturer who has this built-in. I recall 3VR claiming this a long time ago but when I later followed up with them, I believe it was never released into production.

Hi John,

Thanks - I could not find any reference to such a feature in any off-the-shelf VMS out there.

It could be that this is a functional requirement from the end-user and that they are not basing this requirement on any existing platform.

There might be someone out there that does. It certainly can be done. It's not science fiction. It's just uncommon for it to be implemented in production.

Maybe this software would work? Besides blurring, it also can track and label suspects, see below. Also offered as a ftp upload service...

Hi Roger,

Thanks for this, we'll take a close look at how easy this is to do

Cheers

If the product seems reasonable and you get pricing please let us know.

Thanks, Tony

If you want to blur everything what's moving (cars, people, ..), that we've simple app for that, check www.angelcam.com/apps/anonymizer. The app is currently in beta version, provided for free to our current customers to get a feedback what's needs to be done.

I would be happy to hear your specific request so we can add them to the next version release.

Peter, the question here is what product can blue exported video. How can you use your app to do that? Please answer specifically to the request.

John, the bluring is available only for live streaming currently. But as I wrote, we're collecting the feedback (feature requests) for next iteration. Adding a support for exported video (so the video source will not be the RTSP stream from NVR/camera, but uploaded .mp4 file) seems to be one of the favorite. But still, I would like to investigate the particular use cases with more prospective customers to understand the market needs really well.

"I would like to investigate the particular use cases with more prospective customers to understand the market needs really well."

Good for you but that does not solve the problem that the people in this post are asking about.

Make sure to make this clear up front.

OK, will do better next time, thanks :)

At the risk of self promotion, if you need any advice on this or want a solution to take in video post event from various sources, search it, storyboard, annotate etc etc with Kinesense LE. See http://www.kinesense-vca.com/tech/videos/#Report

We also offer services if you do not want to invest in the solution yourself

Sarah, how does your offering solve the poster's need for blurring video? If it does, make that clear asap.

It sounds very promotional and unhelpful. Please clarify immediately or I will delete your comment.

Hi John, all,

For further clarification, there are a few things you need to take into account:

1. Video source: as we all know there are thousands of different video sources out there and often the initial requirement is to get the video into a viewable format without dropping frames. So you need to make sure that if you are converting that you have something that can convert from different formats and you need to ensure that you do not drop frames.

2. Related to the first comment, you need to make sure that you are using a solution that has a complete audit trail so that you can authenticate the processes that you undertook to alter the video (you are changing it and as mentioned in an earlier post, keeping an original copy of the video and working off an a "working copy" is advised.

3. To apply various annotation features such as blurring/ complete masking etc can be a time consuming process and a lot of solutions require you to do this process frame by frame. What we would advocate is that you use a solution that allows you to track a moving object e.g. a person entering and existing the scene, apply the appropriate blur/ masking features using key frames (rather than frame by frame). It speeds up the process. Furthermore, if you have multiple objects in the same scene that you are trying to blur you need to be able to "lock" and the edit you have just done and then progress with masking/blurring other objects without affecting the one you have just done.

If you need further clarification let me know.

Sarah

Sarah,

Please describe the process used in your software to blur out the faces of innocent people in a stream of video (while keeping some faces clear). I went onto your website and did not see much in the way of descriptive information on the process of doing this.

I have tried a few different video editing software in the last week and they are ALL incredibly tedious to learn and to apply filters/masks to faces. Some are frame-by-frame (imagine the work on a 5-minute clip!) and some will follow the movement (Adobe Premiere Pro CC seems to be the best/easiest at this but is still not very easy/intuitive) but none are very easy.

Annotation functionality (bluring, highlighting, adding text etc) is part of our Kinesense LE solution. Essentially once the video is imported into the Kinesense LE solution, it is displayed on a timeline. From the timeline you and scroll and select the clip you want to annotate (for example you might have 10 hours of video and only want to select 2 clips of 2 minutes to annotate). The selected clip is sent to the report tab when you can draw a circle/box around the object you want to blur (or invert the blur to exclude the rest of the scene except the object). You then apply the blur etc to the moving object using key frames. By using the key frames you can do the annotation in seconds rather than minutes. Of course the length of time it takes depends on perspective, what effect you are trying to achieve and the no. of objects you are trying to annotate but generally people find it quick and easy to use.

I am afraid I do not have a video showing the full process rather just the results but I have a screen shot of the annotation tabe which i have included below as a guide.

http://www.kinesense-vca.com/tech/videos/#Report

Normally we provide online one to one demos with customers so that we can understand their complete requirements. I would be apply to arrange this.

Kinesense annotation

When I worked in the Transportation vertical, I cam across a product from Amped Five software that could forensically enhance video, including masking a face or object after the fact. It can also de-blurr video, adjust backlighting and other usefull features. I lost track of them about 2 years ago but here is their web site.

.