This helps us to detect the number on the number plate of a vehicle..
Here for locating the boards I used two different approaches:-
-
DEEP LEARNING + OPENCV Approach - Trained VGG16 model to locate the Number Plate Location with the topx,topy,bottomx and bottomy locations of each image in the Training dataset.
-
OPENCV Approach - Converted Images from RGB to Grayscale, Detected edges and contours.
After detecting the Board Locations by both the approaches I cropped the Number Plate portion and used Pytesseract Image to String and detected the Number on the Number Plate..
Here are the different notebooks:
Deep Learning + OPENCV Approach: Used CNN + OPENCV here for the Vehicle number Detection.
OPENCV Approach: Used purely OPENCV here for the Vehicle number Detection.
I had used the Vehicle Number Plate Detection dataset from kaggle for the Deep Learning + OPENCV Approach
Dataset can be found here.
1. CNN VGG16
2. OPENCV
3. Tensorflow
4. Keras
5. Pytesseract
These were the Car Numbers detected after both the approaches..
We can increase the accuracy of the model by training it with more images with perfect orientation and improved quality of images..
Rahul Kumar Patro