Skip to content

Latest commit

 

History

History
63 lines (36 loc) · 1.86 KB

README.md

File metadata and controls

63 lines (36 loc) · 1.86 KB

Vehicle-Number-Plate-Detection

This helps us to detect the number on the number plate of a vehicle..

Approach

Here for locating the boards I used two different approaches:-

  1. 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.

  2. 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..

Different Notebooks

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.

Dataset Used

I had used the Vehicle Number Plate Detection dataset from kaggle for the Deep Learning + OPENCV Approach

Dataset can be found here.

Technologies Used

1. CNN VGG16

2. OPENCV

3. Tensorflow

4. Keras

5. Pytesseract

Results

These were the Car Numbers detected after both the approaches..

Using Deep Learning+ OPENCV Approach-


Using Purely OPENCV Approach-


Improvement

We can increase the accuracy of the model by training it with more images with perfect orientation and improved quality of images..

Author

Rahul Kumar Patro