Car-Number-Plate-Detector

1. This Project is about a Number plate Recognizer/Reader that could detect the License plate from Car images clicked in an unconstrained condition like occlusion and other spatial variations and read out the Number plate.

2. Extracted Number plate from Car image by smoothing the appearance and then detecting its edges.

3. This was done using the Bilateral filter and Canny edges function of OpenCV.

4. Used computer vision techniques like thresholding, Morphological operations for correcting the perspective transform of plate and segmentation of characters.

5. Trained a CNN architecture model on the character set used in the number plate and got an accuracy of 94% on training data.

Tech Stack

• Python

• OpenCV

• Tensorflow

• Keras.

Sample Plate

VID_20210420_101959.mp4