These python scripts are used to multiply the data from the VOC2012 dataset.
To generate the data, simply create the text file using write_text.py to create the batches of the dataset. Then run the dataset_gen.py to create the dataset.
Please check the path in the files before running the script
To run write_text.py, simply run the command 'python write_text.py '. is the number of the data per file you want. Ex: 'python write_text.py 500' will generate text files each contain 500 data.
To run dataset_gen.py, simply run the command 'python dataset_gen.py --text '. is the name of the file that write_text.py generated. Ex: 'python dataset_gen.py --text 1' will generate a numpy data from text file 1
The dataset_gen.py will generate 3 outputs of numpy array: image data, bounding box data, and class data. Image data will be in format (n, 3, 200, 200) Bounding box data will be in format (n, 50, 4) Class data will be in format (n, 50)
Class data: 0: no data 1: person 2: bird 3: cat 4: cow 5: dog 6: horse 7: sheep 8: aeroplane 9: bicycle 10: boat 11: bus 12: car 13: motorbike 14: train 15: bottle 16: chair 17: dining table 18: potted plant 19: sofa 20: tv/monitor