Searching parking mode footage

[joe384]

Came back to my car to find someone/something has dinged the door quite badly.

As some of you in the A129 forum know, I've been having issues with my camera so I've had the rear camera disconnected for a while.

I do however have the 8 hours of my car parking (low bit rate) from the front camera, so I am hoping I will find out what happened.

Can anyone recommend a utility which will search through the footage for movement or sound spikes?

 

[reverend]

Maybe try something like DVR-Scan

DVR-Scan

dvr-scan.readthedocs.io

If the video has audio then you could extract the audio track and load it into something like Audacity and then look for the spikes in the waveform there and then check the video at that timestamp.

 

[Nigel]

Audacity has a "Sound Finder" on the "Analyze" menu, it will tag noise events for you to review by clicking on the tags.

 

[joe384]

Yeah, I was thinking something like audacity would have it.

Probably need to combine all the videos first and then extract the audio track.

 

[reverend]

Audacity seems it can open MP4 files directly so that's one less job.

Maybe use Registrator Viewer to join it all together first - alternatively the DVR-Scan software should be able to be used with a wildcard on the filename.

Maybe ideas here for @traveler to think about implementing to his software - I've seen a few packages lately pulling in OpenCV support for stuff like this.

 

[Nigel]

Yeah, I was thinking something like audacity would have it.

Probably need to combine all the videos first and then extract the audio track.

Not necessary, although probably easier if you have a way to do it.

You can drag and drop all the mp4 files into Audacity in one go, use alignment end to end on the track menu and then use the Sound Finder.

 

[TonyM]

Not necessary, although probably easier if you have a way to do it.

You can drag and drop all the mp4 files into Audacity in one go, use alignment end to end on the track menu and then use the Sound Finder.

It would be useful to have an example of how this is done, posted on DCT to help users find audio events in parking mode recordings.

I'll give it a go sometime to see how sensitive it is to pick up things like doors opening.

 

[Nigel]

It would be useful to have an example of how this is done, posted on DCT to help users find audio events in parking mode recordings.

I'll give it a go sometime to see how sensitive it is to pick up things like doors opening.

If you have a quiet car and someone dings the door with a trolly, Audacity will easily detect it in A129 Pro audio, not sure about the others, the A129 Pro has the best audio quality. A door opening should be no problem, if you adjust the sliders correctly - not necessarily in the direction you might expect!

I guess someone should do a screen capture video, and one with a real example would be very nice if Joe would like to have a go?

 

[joe384]

I already had a copy of ffmpeg installed so I joined all the files into one big 8 hour file. (Use google)
The sound is aac so to load into audacity you will need ffmpeg library avformat-55.dll (Again just search instructions)

I couldn't figure out the sound finder, it wasn't really needed, I just checked the waveform.



The bit circled above was just an odd tapping noise. I then zoomed into around the 6:30 mark.
Most of it is just rain. The bit circled is the impact noise.



My heart rate has just stepped up a bit. Definitely sounds like a door hitting mine as there is a bounce sound and then a softer thud like a door closing. It's actually painful to listen to.
If you want to listen I have zipped up the wav file so I could attach it to this post.
Can't believe I didn't have the rear camera connected although too be honest that might have caused problems stopping it from recording.

Can't see anyone on the camera so I guess they could have approached from behind? Going to review the footage some more and then try to find the car responsible on monday although it doesn't look like I have any concrete proof.

One thing is for certain... The microphone will pickup most stuff I think.

 

[SawMaster]

I just spent over 2 hours checking vids this morning for a neighbor whose car was stolen last night. Gave the Police what I had, wish it were better but there's a fair chance they will catch the thief as the truck they used isn't common and there's some details in the vids to help.

Phil

 

[TonyM]

I already had a copy of ffmpeg installed so I joined all the files into one big 8 hour file. (Use google)
The sound is aac so to load into audacity you will need ffmpeg library avformat-55.dll (Again just search instructions)

I couldn't figure out the sound finder, it wasn't really needed, I just checked the waveform.

View attachment 48819

The bit circled above was just an odd tapping noise. I then zoomed into around the 6:30 mark.
Most of it is just rain. The bit circled is the impact noise.

View attachment 48820

My heart rate has just stepped up a bit. Definitely sounds like a door hitting mine as there is a bounce sound and then a softer thud like a door closing. It's actually painful to listen to.
If you want to listen I have zipped up the wav file so I could attach it to this post.
Can't believe I didn't have the rear camera connected although too be honest that might have caused problems stopping it from recording.

Can't see anyone on the camera so I guess they could have approached from behind? Going to review the footage some more and then try to find the car responsible on monday although it doesn't look like I have any concrete proof.

One thing is for certain... The microphone will pickup most stuff I think.

Good to know that Audacity helped to find the incident, even if there's nothing to see on the video.

If you really want to capture door impacts on the side, then you need side facing cameras. Although front and rear cameras should pick up most cases of someone arriving or leaving.

 

[Nigel]

I just spent over 2 hours checking vids this morning for a neighbor whose car was stolen last night. Gave the Police what I had, wish it were better but there's a fair chance they will catch the thief as the truck they used isn't common and there's some details in the vids to help.

Phil

I can't think of any way to speed up your search other than playing through at high speed, for me MPC BE plays quite nicely at up to 4x normal speed, and even manages to sort of play the sound at speed, although the sound wasn't much use in your case. Going much faster tends to miss things, but you can speed things up to any speed in a video editor. Of course having side cameras would make things much faster since the car would have been in view until it was stolen, so a quick binary search would easily find the time within a minute.

 

[joe384]

Good to know that Audacity helped to find the incident, even if there's nothing to see on the video.

If you really want to capture door impacts on the side, then you need side facing cameras. Although front and rear cameras should pick up most cases of someone arriving or leaving.

Yeah, there is a car parked to the left and in front, so the culprit must have reversed out. So wish I had the rear camera connected. 😡
I normally park 5 or 6 spaces to the right, but there was a random car there that day. There were a ton of free spaces in front of me I don't understand why they had to park next to me 😡

I have a couple of spare mobius cameras, I am seriously thinking of wiring them up, I guess I will definetly need a dashcam battery pack as well.
Used to think people with more than 2 cameras were mad, I don't now!

 

[TonyM]

Used to think people with more than 2 cameras were mad, I don't now!

Welcome to the club 😀

 

[TonyM]

I have a couple of spare mobius cameras, I am seriously thinking of wiring them up, I guess I will definetly need a dashcam battery pack as well.

A dual port USB powerbank in the boot should be enough to run two Mobius cameras.

 

[joe384]

A dual port USB powerbank in the boot should be enough to run two Mobius cameras.

Well, I have a powerbank for the phone that never gets any use, so I could start straight away getting them setup.
Sorry if you have answered before, but is this what you do? How do you go about charging it, just do it manually?

I could charge up the power bank each night and then just hook up the cameras when I park up at work.

 

[TonyM]

Well, I have a powerbank for the phone that never gets any use, so I could start straight away getting them setup.
Sorry if you have answered before, but is this what you do? How do you go about charging it, just do it manually?

I could charge up the power bank each night and then just hook up the cameras when I park up at work.

I have a 12v socket in my car boot, to power my side cameras and a rear camera when I'm driving. I keep a powerbank in the boot and plug in the side cameras on the occasion that I want to use them.

I have two powerbanks, so I can keep one running while I charge the other one in the office or at home. I get 2-3 days use out of a single charge.

 

[joe384]

Thanks, just read your posts in the side view cam thread. The 2 mobius cameras I have are C2 lenses, I have a 3 door mk7 fiesta with fixed rear quarter windows.
It has curtain airbags so I'll need to be careful with placement. I'll try and dig out my OTG cable so I can test placement tonight.
The 180 fov picture you posted looks good, I will start off with the c2 lenses and go from there.

For today I have hooked one up to a power bank in the boot and sat it on the parcel shelf / rear seat pointing out the rear quarter passenger side.

I have also been reviewing the footage from friday at 8x speed to see anyone crosses across the front of the car.

 

[joe384]

Just happened to be looking at dvr scan, it looks perfect for searching for motion. Something like this:

dvr-scan -i some_video.mp4 -o some_video_motion_only.avi -t 0.15 -tp 30 -tb 30

and then adjusting t (threshold) so it catches cars and people and not branches and leaves.

I think I will have a play around tonight.

 

[joe384]

Hmm dvr-scan seems really slow, in about 20mins of running I cancelled it and it had only picked out a few minutes of movement. (4770K i7)

There must be faster ways out there or maybe something is wrong with my setup. To be continued...