Dash - a new website to view and analyse dash cam videos

zooming a 4k file on a 1080p or lower monitor doesn't match the results you get with a 4k monitor, people that are buying 4k cameras though are more likely to have monitors to suit, if you're going for a picture in picture view along with mapping etc then all video is going to be compressed so some full screen/zoom options are going to be useful anyway
Good point.

By the way, can I ask you all what type of dash cams you have (Make + Model). Would displaying the dash cam the footage was captured from relevant? I am thinking in the case where drivers use multiple dash cams or just want to see this information...
 
I have Viofo and Nextbase.
 
By the way, can I ask you all what type of dash cams you have (Make + Model).

In alphabet order (all 1920 * 1080p 30fps unless stated otherwise):

o Blacksys CF100
o Cobra 840 (sometimes set for 720p 60fps)
JooVuu X (2560 * 1080p, 30fps)
o Mio Mivue 528
o Mio Mivue 538
Mobius C2
Panorama II S
Street Guardian SG9665XS V2
Street Guardian SGZC12RC V2
o Transcend DP100
o Viofo A118C

Those prefixed with 'o' are older cams which don't see regular use but are sometimes used as spares or for experiments.
 
Last edited:
I have Viofo and Nextbase.

Thanks Paul. We are actually currently testing metadata extraction on both Viofo and Nextbase so hopefully we will have these covered :).

In alphabet order (all 1920 * 1080p 30fps unless stated otherwise):

o Blacksys CF100
o Cobra 840 (sometimes set for 720p 60fps)
JooVuu X (2560 * 1080p, 30fps)
o Mio Mivue 528
o Mio Mivue 538
Mobius C2
Panorama II S
Street Guardian SG9665XS V2
Street Guardian SGZC12RC V2
o Transcend DP100
o Viofo A118C

Those prefixed with 'o' are older cams which don't see regular use but are sometimes used as spares or for experiments.

Thanks for the info! Can I ask you why you have so many dash cams? Is because you test them, have multiple vehicles, or any other reason?
 
I help with testing, yes. There are members who do more, and test other features than I do, but dashcam companies benefit from people helping them in the early days prior to release.
 
Thanks for the info! Can I ask you why you have so many dash cams? Is because you test them, have multiple vehicles, or any other reason?

Our household has multiple cars (three or four) and each car is usually fitted with two front and one rear cam. Two front cams greatly reduces the chance of not capturing an important moment due to a problem with a cam, its memory card, power supply or user error. I have two or more of some of the cams listed.
I have done some dashcam testing in the past.
 
Our household has multiple cars (three or four) and each car is usually fitted with two front and one rear cam. Two front cams greatly reduces the chance of not capturing an important moment due to a problem with a cam, its memory card, power supply or user error. I have two or more of some of the cams listed.
I have done some dashcam testing in the past.

Wow that's quite comprehensive! I am learning so much here on the different use cases for dash cams. That's pretty creative what you do! Would it be important for you then to identify which camera and car the footage is from then? Apart from you login into Dash, do you think another member of your family may want their account? It feels we would be more moving towards fleet management with this approach but it could be suitable for private drivers too.

One thing I have been wondering as well is whether beyond video and showing location on a map drivers would be interested in some kind of reports. Do you get statistics like how many miles/km you drove or things like average speed across your journeys from your current dash cams or associated software?
 
One thing I have been wondering as well is whether beyond video and showing location on a map drivers would be interested in some kind of reports. Do you get statistics like how many miles/km you drove or things like average speed across your journeys from your current dash cams or associated software?
Hard to think of any statistics that would actually be useful to a normal person, and to get them we would presumably have to upload everything. For fleet operations it may be possible to reduce fuel use and may be possible to distinguish good driving from bad, but those statistics are much easier to obtain by watching how much fuel you put in the vehicles and how many repairs/accidents/complaints you receive about a driver.

I've got four Viofo cameras, A119 V3, A129 IR, A129 Pro and MT1 (Dual motorcycle camera), plus two Blueskysea cameras, B1W and B2W, plus a few older cameras I don't use anymore. While Paul "helps with testing" to remove the bugs, I try to help remove the mistakes before they ever get built.
 
I'm all for anything new and good inre dashcams, but I'm wondering what is offered here other than storage that isn't available through other viewing platforms?

Competition is good but when the market is small, competition also spreads that limited revenue out reducing everyone's ability to better develop their products.

Still liking this even though I'm not sure it will be useful for me.

Phil
 
I'm all for anything new and good inre dashcams, but I'm wondering what is offered here other than storage that isn't available through other viewing platforms?

Competition is good but when the market is small, competition also spreads that limited revenue out reducing everyone's ability to better develop their products.

Still liking this even though I'm not sure it will be useful for me.

Phil

Hi Phil,

Thanks for your feedback.
Besides online storage, we also enable you to view your videos together with position and speed data in an interactive way, like some applications out there. But that is really just the start for us. We want to learn more about what product we should build for the community, and there are some very interesting suggestions out here such as zooming, sound analysis etc. We are definitely going to have some computer vision around detection of other cars and number plate, etc., together with some event identification (e.g. collision, near miss, etc.) and explanation.

I'd be interested in learning more about the kind of product you believe is missing in the market :).

Thanks,
Ed.
 
not everyone wants or needs the online storage side of things, have you thought about doing a player that can just be used locally, currently the only solution in the market (outside of any cameras that include a player) is a USD $25 option
 
I'd be interested in learning more about the kind of product you believe is missing in the market :)

One thing I feel is highly needed is a video enhancer; something that could help folks zoom in on and retrieve details such as number plates which may not be clear when viewed normally. It can be done DIY now but it takes lots of time to do each step frame-by-frame. That would be lots easier if the zoom, contrast, brightness, and colorization controls were on screen and linked together automatically with some AI programming attempting the job, or at least synchronizing those controls to give what should be best results by using one control. Then the users would only need to slightly fine-tune from there. It's a huge job I know and not something which could be made perfect every time but it's something we're currently lacking and eminently useful for nearly all of us. Perhaps offer it with number-of-use restrictions then have a 'purchase' option if the user likes what they see and wants more. You're going to have to pay for all this somehow!

Phil
 
That would be lots easier if the zoom, contrast, brightness, and colorization controls were on screen and linked together automatically with some AI programming attempting the job,
Add in an easy to use slowmo that repeats a few frames with the zoom and enhancements to make it easier to work out what happened.

And motion deblur. Motion blur is probably the biggest cause of not being able to read plates and it is hard to find any software that will easily do a decent deblur which is aimed at dashcam use. With a dashcam, you know the blur direction always follows the perspective lines from the point on the image where the car is heading and increases predictably towards the corners of the image, so once you locate the centre of the image for the perspective, not necessarily the centre of the actual image since the camera may not be facing directly forwards, then removing the motion blur should be easy to do and effective. Of course sometimes you also have to allow for your car turning which may add an extra horizontal motion across the frame.
 
not everyone wants or needs the online storage side of things, have you thought about doing a player that can just be used locally, currently the only solution in the market (outside of any cameras that include a player) is a USD $25 option

We thought about this, but to be fair there are already a number of options available out there for desktop players. I just didn't feel we could offer much more than what the other software do in this domain.

One thing I feel is highly needed is a video enhancer; something that could help folks zoom in on and retrieve details such as number plates which may not be clear when viewed normally. It can be done DIY now but it takes lots of time to do each step frame-by-frame. That would be lots easier if the zoom, contrast, brightness, and colorization controls were on screen and linked together automatically with some AI programming attempting the job, or at least synchronizing those controls to give what should be best results by using one control. Then the users would only need to slightly fine-tune from there. It's a huge job I know and not something which could be made perfect every time but it's something we're currently lacking and eminently useful for nearly all of us. Perhaps offer it with number-of-use restrictions then have a 'purchase' option if the user likes what they see and wants more. You're going to have to pay for all this somehow!

Phil


Add in an easy to use slowmo that repeats a few frames with the zoom and enhancements to make it easier to work out what happened.

And motion deblur. Motion blur is probably the biggest cause of not being able to read plates and it is hard to find any software that will easily do a decent deblur which is aimed at dashcam use. With a dashcam, you know the blur direction always follows the perspective lines from the point on the image where the car is heading and increases predictably towards the corners of the image, so once you locate the centre of the image for the perspective, not necessarily the centre of the actual image since the camera may not be facing directly forwards, then removing the motion blur should be easy to do and effective. Of course sometimes you also have to allow for your car turning which may add an extra horizontal motion across the frame.

Duly noted! These are technically challenging features to ship, but I can see how they could greatly enhance the value of the product. We are definitely going to do some R&D into some of these areas after we have launched and polished Dash. Thanks for all the feedback!
 
Hi everyone,

Just to let you know that we are still working on Dash :D! I wanted to share with you some of the little progress we made about a week ago. We are now able to sync video playback with GPS position and speed (which we both extracted from my Viofo A119 V3). This is a small video to show what we are able to do now:


Would love to have your thoughts/feedback on this.
 
Does anyone have a sample footage from a dash cam other than:
* Viofo
* Nextbase
* Garmin

We are looking to start adding more compatibility among the dash cams we support. At the moment we are focusing on single channel dash cams. Dual channel ones will come soon. You can PM me to share this information.

Thanks so much!
 
At the moment we are focusing on single channel dash cams. Dual channel ones will come soon.
Don't forget the multi-channel cameras!

Most of the dual, triple, etc channel cameras will have seperate video files for each channel with each one containing an exact copy of the sound and GPS data, so it should be easy to include them if you allow for that in the design from the start.

More difficult are the ones that don't use .mp4 files, but currently there are not many decent dashcams in that category, although I think your Viofo can be set to use .ts files, you might want to cater for them, especially since it is hard to find any player that will read the GPS data out of .ts files...
 
Don't forget the multi-channel cameras!

Most of the dual, triple, etc channel cameras will have seperate video files for each channel with each one containing an exact copy of the sound and GPS data, so it should be easy to include them if you allow for that in the design from the start.

More difficult are the ones that don't use .mp4 files, but currently there are not many decent dashcams in that category, although I think your Viofo can be set to use .ts files, you might want to cater for them, especially since it is hard to find any player that will read the GPS data out of .ts files...

We are certainly not forgetting multi channel cameras ;). Although I presume dual would be the most common type?
I think it should be easy easy-ish to process videos from multi channel cameras if the metadata is almost the same. I wonder though if within the metadata there is information about whether the video is from the front/inside/back facing camera. Otherwise it becomes trickier to figure this out on our own.

I am somewhat familiar with .ts format because, we actually transcode the original videos to HLS to make them viewable online and adjust quality depending on your device's connectivity, etc.
 
I wonder though if within the metadata there is information about whether the video is from the front/inside/back facing camera. Otherwise it becomes trickier to figure this out on our own.
typically the file names will have different name format, they'll have A, B, or F, R designation in the files
 
We are certainly not forgetting multi channel cameras ;). Although I presume dual would be the most common type?
Currently single is probably most common with dual not far behind (some front+rear and some front+interior). I think once triple cameras become more available at reasonable cost then they will become very popular, why would a taxi choose between a rear or an interior camera, much better to have front+rear+interior. Even for a private car the interior is useful, it shows that you were not on the phone when you crashed even if there is a witness claiming that you were, and catches cars actually hitting your side instead of just assuming that the collision noise came from the car that just disappeared from the view of the rear camera. An interior camera is easy to fit since most will be built into the main camera housing, and if not then it will only be a short cable to run so there is no reason not to get one other than a small extra cost and some people being camera shy. For company vehicles, it would allow you to analyse the videos to check that drivers are paying attention to the road at all times, and not using phones or falling asleep.

I wonder though if within the metadata there is information about whether the video is from the front/inside/back facing camera. Otherwise it becomes trickier to figure this out on our own.
Like jokiin says, currently the more popular cameras distinguish them by file name, with each brand using a different file name format, and in the case of Viofo each model using a different file name layout! But they normally end with F, R and I, for Front, Rear and Interior, maybe C for Cabin, if the manufacturer bothers to change from the default A, B, C of Novetek based cameras. Of course the dual taxi cameras often use F and R for Front and Interior!

I guess it would be good to include a standard attribute in the video stream attributes of the MediaInfo (allows the possibility of multiple streams in one file).
View : Interior
 
Back
Top