Close

New Camera, New Take on the Bicycle Dashcam

A project log for Bicycle Dashcam

Bicycle Dashcam with License Plate Recognition (ALPR)

Richard AudetteRichard Audette 04/25/2021 at 01:444 Comments

I started a fresh design with a different idea, a new camera, new models, and new code, which I submitted as an entry for the Toronto ♥️’s Bikes Make-a-Thon.

This version of my dashcam has blind spot detection, similar to what you would see in modern cars.  I'm now using a USB AI accelerated camera from Luxonis mounted to my bicycle seat post, and my smartphone as a display. It’s a few hundred lines of Python code that builds on a freely available AI vehicle recognition model from the Intel Open Model Zoo. I’ve built on the license plate recognition and MJPEG video streaming sample code from Luxonis that was supplied with the OAK-D camera. I tether the laptop (I hit a snag I didn't have time to address with my Pi) to the smartphone using wifi, and I use an iOS app called IPCams to view the video stream. 

The vehicles are recognized and identified. The video is streamed over wifi to the smartphone. A caution alert is added to the video when a vehicle is detected.



Demo video:

This camera mount worked well for me:
https://www.aliexpress.com/item/32795104954.html

And I used this app to view the MJPEG stream on the phone:
https://apps.apple.com/us/app/ipcams-ip-camera-viewer/id1045600272

And... you can download the source for this iteration here:
https://github.com/raudette/SmartDashcamForBikes



Discussions

Brandon wrote 09/22/2021 at 20:51 point

Hi Richard,

Great, thanks!  This may be a duplicate as somehow my draft got deteted...  But here goes.

So we are making it so OAK-D-Lite backers can choose fixed focus.  Which will help a TON with the vibration.  And we're also making an OAK-D variant with fixed focus.  And... we're working to improve our OCR flow.  Which will help a lot here too.

Thanks again,

Brandon

  Are you sure? yes | no

Brandon wrote 09/22/2021 at 20:03 point

Thanks @Richard Audette !  All sounds great.  So we will also be offering the OAK-D-Lite in fixed-focus now as well.  Which should help.  And we have some exciting advancements on OCR.

  Are you sure? yes | no

Brandon wrote 09/17/2021 at 03:01 point

Hi Richard,

I don't know how I missed your post here.  But nonetheless, I just found it!  We launched a new KickStarter yesterday for a lower-cost version.  So it won't be quite as expensive!  

https://www.kickstarter.com/projects/opencv/opencv-ai-kit-oak-depth-camera-4k-cv-edge-object-detection/description

Also, we have in the works a version that can run standalone and give results over Bluetooth and WiFi.  It has a nice enclosure as well.  Just add power, and it's good. 

If you'd like to try it out - we'd be more than happen to send you one of the first prototypes when they're available in about 2 months.  Feel free to shoot me an email at brandon at luxonis dot com.

Oh also we're working on fixed-focus RGB... to handle vibrations a TON better.

Thanks again,

Brandon

  Are you sure? yes | no

Richard Audette wrote 09/22/2021 at 00:46 point

Hi Brandon - I'll reach out. A fixed-focus RGB camera would be great for this use case - I've had some success in tweaking this camera's properties to accommodate bumpy rides. 

Since this post, I've been working on a few things (at a hobbyist's pace!):
- Collecting some sample imagery on rides using https://github.com/luxonis/depthai-experiments/tree/master/gen2-record-replay
- Looking specifically at clarity of license plates, the best results I'm getting so far is reducing to 2 fps (can improve - a limitation of the Pi 4/thumb drive I'm using in development), recording at 4056x3040, auto focus off, locking the focus at 120, scene mode sports
- Learning how to train my own Yolov3 model with local license plates

Congrats on the Oak-D Lite launch - that price point will certainly make new use cases possible.

  Are you sure? yes | no