ekorudiawan / Playing-Card-Classifier

Classical Computer Vision with Machine Learning Pipeline to Identify Playing Card

Playing-Card-Classifier

Dependencies

• Python 3.7
• OpenCV 3.4.6
• scikit-learn

List Source Code

• ./sources/Image_Augmentation_Tool/Image_Augmentation_Tool.py : Source code for generating syntetic dataset from original image
• ./sources/HOG_Classifier/HOG_Classifier.py : Source code for classify the image using HOG feature detector
• ./sources/ORB_Classifier/ORB_Classifier.py : Source code for recognize the image using ORB feature detector

List Dataset

• ./original_images : This is a directory that contains original image of 52 type playing card
• ./dataset/train : This is a directory that contains train images that was generated by source code Image_Augmentation_Tool.py
• ./dataset/test : This is a directory that contains test images that was generated by source code Image_Augmentation_Tool.py

How to Run The Program

Generating Synthetic Image

cd ./sources/Image_Augmentation_Tool
python Image_Augmentation_Tool.py

Run Card Detector using HOG Feature

cd ./sources/HOG_Classifier
python HOG_Classifier.py

Run Card Detector using ORB Feature

cd ./sources/ORB_Classifier
python ORB_Classifier.py

Notes

Extracting feature, generating train dataset and training the classifier is taking around 10 minutes. After classifier successfully trained, program will predict all images in test dataset folder. If there is a wrong classification, program will show up the image. To continue classification process, just exit the image window. Program will result in classification result at the end.