gabrock94 / pyaesthetics Goto Github PK

Hi @Gabrock94, thank you for the awesome repository.

Looing at this code, the implementation appears to be in progress. Are there any plans for future implementation?

• Fix warning in text / image ratio when area of the image == 0
  /usr/local/lib/python3.5/dist-packages/prettywebsite-0.0.4-py3.5.egg/prettywebsite/spaceBasedDecomposition.py:63: RuntimeWarning: divide by zero encountered in long_scalars return({"textImageRatio":sum(text) / sum(image),"textArea":sum(text),"imageArea":sum(image),"nImages":len(image)})

At the moment, one of the bottleneck of the package is the detection of the ratio between visual and textual areas.
The process involves the detection of areas by means of a Space Based Decomposition, followed by an OCR detection for each area. The module should be rewritten to avoid the need of running the OCR algorithm several times such as in https://www.geeksforgeeks.org/text-localization-detection-and-recognition-using-pytesseract/, and then compare empirically the areas detected via the space based decomposition