speed camera Runs fine on a RPI-3B at default resolution 320x240 or 640x480. RPI5 would be overkill but I suspect you could get higher stream resolution. I would use an IP/RTSP security camera stream so the RPI can be placed anywhere on your network, IP cam can be mounted outdoors on wifi or wired. RPI5 could run other things on it with speed-cam. I use Topodome security cams. They are pretty cheap on Amazon. Claude

Lots of people have installed this setup and contributed to project. Easiest way is to access the github page at github.com/pageauc/speed-camera. There is a wiki that provides additional help. Easiest install is using a RPI computer (any model will work) with a USB camera, Raspberry Pi cam or RTSP security camera. Install software using curl command per github page instructions. This assumes you are comfortable with putty SSH and/or UNIX Terminal (Other tutorials cover this) Also this video might be helpful or not kzfaq.info/get/bejne/pJiHh5aCx5a3eYk.html

I don't have an HQ camera so cannot test. I suspect it might be sending a higher resolution than the default 320x240. Run speed-cam.py from terminal session and check the logs eg INFO speed_notify PILIBCAM video stream size is 320 x 240 Log message is a few lines above Begin Motion Tracking ....... It could also be a problem with libcamera python library and its config settings (not sure) The strmpilibcam.py file does include raw setting for libcam setup that a previous user said resolved his HQ cam issue self.picam2.configure(self.picam2.create_preview_configuration( main={"format": 'XRGB8888', "size": self.size}, raw={"size":self.picam2.sensor_resolution}, transform=Transform(vflip=self.vflip, hflip=self.hflip))) I suggest you contact me via github Issues (top menu 2nd item) on the speed-camera repo at github.com/pageauc/speed-camera Claude ...

@@ClaudePageau many thanks for your speedy reply. I will try the streaming settings initially and raise on Github if I cannot resolve.

For a RPI 3 I would not recommend a resolution greater than 640x480 since opencv has to process the stream frames with fast enough FPS to track objects in real time even though the HQ cam is capable of much greater resolutions. Personally I have been using topodome bullet security cameras reading the 640x480 RTSP 12 stream feed. The cameras are cheap on Amazon (much cheaper than HQ cam (approx $40 USD) . They are waterproof, have wi-fi or rj45. Can be mounted outside within Wi-Fi/RJ45 range and pointed towards road. Can also be used as a security cam concurrently. The RPI running speed-cam can then be placed almost anywhere inside on your LAN nice and cozy and safe. You just need to config cam with fixed IP through the camera web ONVIF interface or via DHCP on router. I also use free security camera software at ispyconnect.com running on a hub computer. I use RPI4 as my hub for 5 cameras. Two cams feed to RPI running speed-camera. I subscribe to view cameras remotely (no port forwarding) via device app (Eyespyconnect Agent DVR Client on Android Play or IOS App stores) or runs on a device or PC web browser . Works great for me. Note You don't need ispyconnect if you only want to use bullet cam for speed-camera. Just wanted to let you know the dual purpose solution I use. Regards Claude ...

Code is written in python. You can see code, wiki, Readme.md on the speed-camera repo per link in KZfaq Description. This video was done a long time ago and was done as an early demo. I am retired and after 500+ videos I have taken a break from uploading KZfaq content. Yes it can run on MAC, Windows and other varieties of Linux distros (curl install should work on any Debian version. Can use USB cam or RTSP IP Camera. Mac and Win may have problems running BASH scripts but I know newer windows 11 supports a Linux Windows Subsystem. You can also run under Docker but not sure how up to date it is since speed-cam.py is now recently at version 13.02 I prefer RPI with a cheap outdoor, low res RTSP stream. IP/RTSP Cam can be located outside with WIFI or wired connection and RPI can be located anywhere in the house since RPI speed images and reports can be viewed via a webserver on any most computer, device web browser on default 8080 port (This can be changed). You can also use rclone to upload images to an online file storage provided. See rclone website for more details. I use my google GDrive since it is free. Max images can be set to manage space. I am not saying all this is easy peezy but wiki instructions can help. Since speed camera uses motion tracking, false positives are greatly reduced. Much better than motion detection that is built into most security cameras. I have used it for security just turn off the speed display. Good Luck. You can contact me on my GitHub Repo if you have an issue.

It was between Wellesley and Tiviotdale Ontario, Canada

This video is not related to speed camera but I will answer your questions. It would be better corresponding on GitHub There is no setting for day only mode. Speed still gets calculated via headlights though. The only way to control would be to create crontab entries to start and stop speed camera using the speed-cam.sh script. eg sudo crontab -e Add the following (edit as required) 0 8 * * * su pi -c "/home/pi/speed-camera/speed-cam.sh start > /dev/nul 2>&1" 0 20 * * * su pi -c "/home/pi/speed-camera/speed-cam.sh stop > /dev/nul 2>&1 ctrl-x y to save and exit This will setup two cron jobs one to start at 8am and one to stop at 8pm Camera would work at an angle but calibration would be unreliable. At one time I was looking at having the speed camera send a signal to a second camera pointing sideways. That could be used for licence plate reading. Never did anything with the idea though. This would work similar to my panopi project on github github.com/pageauc/panopi

In config.py take a look at the Motion Event Setting x_diff_max = 24 This is the max number of pixels allowed between events. With higher resolutions this would need to be larger. Take a look at the logs and see out messages to get an idea for a better setting. This setting filters out motion tracking events that are unreasonably far away from the last event. Default is for 320x240 stream size. Fast objects will easily exceed this value with larger image sizes. You can also take a look at the track_counter setting. Just a reminder that if the camera is too close to the objects to be tracked very fast objects can go by too quickly to be tracked especially with with larger image stream size that slows down opencv processing. Ideally the track time across the image monitoring rectangle should be at least 1-2 seconds so moving camera farther back will give a wider view and increase the track time. Let me know if this helps.

@@ClaudePageau hi Claude - thanks for the response! My camera is approx 40-45 feet away from the street. I don’t really have an option to move it closer/further. I did update on your suggestion and log analysis and was able to use track setting of 8 to some success. I’m still unsure of the sweet spot with track setting and high speeds. My calibration ox and mm seemed to be in line with some images and averages I took. My new troubleshooting is focused on new tracking and captures. If the setting is 0 I will pick up the same vehicle twice but if it’s too high I seem to run the chance to miss the capture. I’m also trying to look at the right size for contour sq pixel capture. Question - in calibration mode does the program ignore the min_area and mph?

For rough positioning move camera to point correctly. You can then use config.py settings per below to set the motion tracking red rectangle per the x and y left, right, upper and lower settings. The settings below are for 320x240 image stream. Try not to make too large since opencv will need more processing for larger rectangles. Just try to cover the essential motion tracking area. # Motion Tracking Window Crop Area Settings # ----------------------------------------- # Note: Values based on 320x240 image stream size. # If variable is commented, value will be set automatically based on image size. # To see motion tracking crop area on images, Set variable image_show_motion_area = True # Set align_cam_on = True to help with adjusting settings. x_left = 50 # Default=50 comment variable for auto calculate x_right = 250 # Defaykt=250 comment variable for auto calculate y_upper = 90 # Default=90 comment variable for auto calculate y_lower = 150 # Defaykt=150 comment variable for auto calculate

@@ClaudePageau I see the red hatched bar high up on my window. The red rectangular box with the green circle is within the hatched bar which goes across the entire screen. All I need to do is lower it so it covers the street and not the trees. config.py shows the following # Motion Tracking Window Crop Area Settings # ----------------------------------------- # Note: Values based on 320x240 image stream size. # If variable is commented, value will be set automatically based on image size. # To see motion tracking crop area on images, Set variable image_show_motion_area = True # Set align_cam_on = True to help with adjusting settings. x_left = 50 # Default=50 comment variable for auto calculate x_right = 250 # Defaykt=250 comment variable for auto calculate y_upper = 90 # Default=90 comment variable for auto calculate y_lower = 150 # Defaykt=150 comment variable for auto calculate

@@ClaudePageau I adjusted the motion tracking window crop area settings. This had no effect on the positioning of the hatched bar/red rectangle/green dot

@@IanNarcisi Hatched bar is the calibration marks where each mark is 10 pixels. This is used to measure the vehicle size in pixels per calibration details in wiki. This can be shown or hidden using the config.py calibrate = True or False. False hides hash calibration marks. Position is calibrated and there is no setting to adjust except to modify the program code.

@@ClaudePageau I understand. I believe my USB 1080p Logitech HD camera is causing this hatching to be seen so high on the web images. I wish I could attach a photo to show you what it looks like. In essence, I would have to aim the camera at the grass below in order to get the calibration marks over the road. Perhaps I just need to order a picam. Thank you Claude.

Danke

I talked to Michael and he said all his snow came from Sault Ste Marie. Delivered by Mother Nature Transportation Company. LOL. So it was all from his yard.

I designed it to run on very old rpis single core and 500mb ram memory so cheaper or older models should do fine. You can also use a cheap ip of USB camera instead of rpi cam

I can run on old raspberry pi's with 500mb ram but usually run Lite raspberry pi OS for better performance.

Can be a USB webcam almost anything will work or IP Camera with RTSP streaming protocol. Check the users manual. I prefer the pi-camera but there will be issues with the new Bullseye OS ver of Raspberry Pi OS. Waiting for the new python picamera2 library

@@ClaudePageau Good to know. Thank you Claude!

@@ClaudePageau Good to know. Thank you Claude!

@@IanNarcisi Your Welcome

Sure. Let me know what you are thinking

So sorry to hear that Claude. She seemed like such an adorable sweetie, and touched many lives through your videos.

@@bonnieluvstheblues6777 Thank You. She was actually almost 15 years old. We had her cremated and Renee keeps the beautiful box of ashes in her sewing room. Renee got a female 4 month old black and tan short haired dachshund pup from Weslaco Texas. Miss Mae was a rescue and we got her at no cost from the woman who was fostering her. Also there was a clinic in Edinburg Texas where she got her Rabies, Microchip and Immune shots Free. Had to wait in line for a long time (4 hours) though. She is up-to-date on all her shots and is very healthy with no problems so far. She is also a very smart/loving/friendly dog. Rosita is managing and it is nice to see them snooze together and chase around the back yard and in the house..

Sorry could not find the image you are looking for. Usually take lots of images but this time just did mostly video clips. Claude

@@ClaudePageau I was just wondering because the car use to be own by a friend of mine

@@williambaker4213 Do you think it might have been entered for other years of Pumpkinfest or Sauble Sandfest events. If so I could look at other years. What year was Dodge Dart (approx) Can't be too many yellow Darts

@@ClaudePageau I don't know because that was the new owner of the car

Crane working on new houses being built