A Dutch plate recognition written in Python using only image processing techniques and no machine learning. The process involves:
- Plate localization
- Character segmentation
- Character recognition
- Adjusting the output
The system works well on videos of Dutch license plates up to about 10m distance.
You run your video by putting under dataset/trainingvideo.avi
.
You can see the output under dataset/Output.csv
.
The shell script evaluator.sh
is used for running the project and calculating scores.
This file initially runs your main.py
file, followed by evaluation.py
.
Do not modify either one of evaluator.sh
or evaluation.py
if you want to see proper outputs.
Rest of the project:
CaptureFrame_Process.py
for reading the input videoLocalization.py
for figuring out the location of the plate in a frameRecognize.py
for figuring out what characters are in a platehelpers/
additional methods to help you get started (you do not have to use them, and are there for inspiration)requirements.txt
if you want to use additional Python packages make sure to add them here as well