giscle / data-preparation

All the scripts related to data preparation.

Data-Preparation

All the scripts related to data preparation.

                                        ********** rename.py *************

This Script is used to Maintain Uniform Annotation Label in XML Files Across the Dataset which is Annotated Using LabelImg Software Tool

For eg : Person X annotated a car as "CAR" person Y annotated a car as "cAR" , then this code will convert all Annotations to "car" irrespective of Perosn X and Y without any Hassle.

To run this Script : Place this rename.py in the Folder which has Image and Annotations together and set the path "arr" to this current Directory and Run

                                     ********** duplicate_remover.py ***********

This Script is Used for Removing Unannotated Data done using LabelImg or LabelMe Tool . In Many Images there may not be objects to annotate , that image will not have its corresponding .XML File , hence it may contribute as a Negative Weight while Training Data. So to prevent that this code is written , which removes all Unannotated Images from the directory .

For eg : A folder contains [ image1.jpg , image1.xml , image2.jpg , image3.jpg , image3.xml] Here image2 is unannotated , so it does not have its corresponding xml . After running this script the directory content is given as follows [ image1.jpg , image1.xml , image3.jpg , image3.xml] , here we see image2.jpg is removed .

To run this script : image_directory stores the cureent directory of this file location , to run a script on a specific folder change 'day more' to the folder name and Run this script

                                      ************* file_renamer.py **************

This script is to rename the Files and serialize them accordingly so as to make the Dataset Clean and Organized for Training

For eg: A folder contains [ image1.jpg , image1.xml , image3.jpg , image3.xml , image11.jpg ,image11.xml ] . Here image2 ,....image10 is missing / removed by duplicate_remover.py script , now this data is not Arranged , if we run this file_renamer.py script on this folder then the folder content is [ image_1.jpg , image_1.xml , image_2.jpg , image_2.xml , image_3.jpg ,image_3.xml ] , so here all data is arranged and renamed in ascending order like image3.jpg got renamed to image_2.jpg and so on

To run this script: image_directory stores the cureent directory of this file location , change 'day more' to the folder name on which it will be applied , and mention from where to start counting by setting the start value to count_n variable and run

                                    ************* Order to Run the Script **************
                                    
                                       1. Run the rename.py on a Directory 
                                       2. Then Run duplicate_remover.py on the Directory
                                       3. Then Run file_renamer.py on the Directory