Python script to count cars by using OpenCV lib optimized for using on a Raspberry Pi 4.
Program counts cars on a video stream and uploads the counts to Thingspeak.com. It differentiates between the direction of travel. Furthermore it uploads the CPU temp of the pi as a part of healthmonitoring of the system.
In Debug Mode (postfix -d) a file output.avi is saved with marked image crop.
The source code in visualize_traffic_counter_24h.m is the visualization code vor Thingspeak.com.
Clone repo to your Raspberry Pi and insert your Thingspeak.com write-key key = "HERE_YOUR_THINGSPEAK_WRITEKEY". Run python3 car_counter_BGsub.py -d first to capture a video output.avi from the camera. After that define the picture crop to resize the detection zone. X1, X2, Y1, Y2 are the corenerstones of the detection zone in px. Eg.:
# Define picture crop -> Check in debug mode!
roixy = [ 470, 675, 200, 337 ] # X1, X2, Y1, Y2
After that rerun python3 car_counter_BGsub.py -d and check the correct place of the detection zone. After all you can run the programm python3 car_counter_BGsub.py to start counting and uploading the counts to Thingspeak.com
I build for the Raspberry Pi and the Camera a waterproof case from a sanitary pipe. So, the system runs now for month without any problems. You can check the results on my Thingspeak.com-channel.
python openCV numpy gpiozero
Inspired by https://www.youtube.com/watch?v=HXDD7-EnGBY