Project to practice OCR and OpenCV. Used a variety of methods to gather different information from driver's license test image
Ran driver's license through EAST Text Detection model to get region's of interest for all the text fields then ran each ROI through pytesseract(OCR) to turn text to string. Ran text strings through a regular expression to grab strings that contain dates.
git clone https://github.com/Gabefire/dl-object-recognition.git
cd dl-object-recognition
python3 text_detection.py --image test-image.jpeg
Trained model on k nearest neighbors data located in the train folder. Evaluated accuracy with images being converted to raw pixels and images being converted to a histogram once model was trained compared it to test-image.jpeg to figure out the most likely state.
git clone https://github.com/Gabefire/dl-object-recognition.git
cd dl-object-recognition
python3 knn_classifier.py -d ./train/ -k 3
Used openCV and pytesseract to find the date fields of the id. Processed image, find contours of text and then ran each contour through OCR to get date fields.
git clone https://github.com/Gabefire/dl-object-recognition.git
cd dl-object-recognition
python3 text_dect_openCV_only.py -i test-image.jpeg