-
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.
-
Extracted Number plate from Car image by smoothing the appearance and then detecting its edges.
-
This was done using the Bilateral filter and Canny edges function of OpenCV.
-
Used computer vision techniques like thresholding, Morphological operations for correcting the perspective transform of plate and segmentation of characters.
-
Trained a CNN architecture model on the character set used in the number plate and got an accuracy of 94% on training data.
-
Python
-
OpenCV
-
Tensorflow
-
Keras.
