xeryustc / handwritingrecog Goto Github PK

Parse all the supplied files to find the possible words and their frequencies.

Check the cutouts to see if characters can be categorized/cutout by using some sort of prototype window size

Combine creation of segments, extracting hog features and training svm or other modules in one script

Complete the pipeline, so that we can correctly classify the words in a given image file by using the given input/output structure,.

Re-organize the code so that it corresponds with the given structure, i.e.:

1. The main file is called recognizer.py and runs the entire pipeline. It requires 3 arguments:
   Input image, Input words file containing the word locations, output words location where we need to write our results.
2. All the code should run after a single make command
3. Our own code should walk through the pipeline by using modules, every step (preprocess, etc.) should be a different module, called from the main file: recognizer.py