Image parser with OpenCV + Tesseract
The purpose of the parser: to extract the information of certain cells and enter the database (excel table)
Complexity: the window position varies on the desktop, and in some cases the window interface changes. OpencCV is used to find interested fields by segmenting the image, filtering the found fields and creating individual images (in the OZU) to be transmitted to the tesseract.
accuracy of recognition: >95%. About 5% of the error is the tesseract "wine", but the standard base was used without training. I am sure that with training you can at least 99%. The accuracy of the cells is 100%.
Comparison with template matching and machine learning is not used. Supervised learning can be used to find all the fields of interest, but for this example it would probably be overlooking.
#OpenCV #tesseract #C++ #Parsing #Win32API
The purpose of the parser: to extract the information of certain cells and enter the database (excel table)
Complexity: the window position varies on the desktop, and in some cases the window interface changes. OpencCV is used to find interested fields by segmenting the image, filtering the found fields and creating individual images (in the OZU) to be transmitted to the tesseract.
accuracy of recognition: >95%. About 5% of the error is the tesseract "wine", but the standard base was used without training. I am sure that with training you can at least 99%. The accuracy of the cells is 100%.
Comparison with template matching and machine learning is not used. Supervised learning can be used to find all the fields of interest, but for this example it would probably be overlooking.
#OpenCV #tesseract #C++ #Parsing #Win32API