Coco Format data parsing for EDA

[SangJunni]

Coco Format parsing for EDA (just paste and use this!)

#import libraries
import json
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib
import numpy as np
import os
from pycocotools.coco import COCO

#image loader
src_root = 'root folder of dataset in your environment'
src_subset = 'subset root of your environment'
src_file = src_root + 'train.json'
src_desc = 'a name (identifier) for the dataset'

coco_obj = COCO(src_file)

#data load
with open(src_file, 'r') as f:
    root = json.load(f)
    
root.keys() #dict_keys(['info', 'licenses', 'images', 'categories', 'annotations'])

#Basic High level infromation
n_images = len(root['images'])
n_boxes = len(root['annotations'])
n_categ = len(root['categories'])

# height, width
heights = [x['height'] for x in root['images']]
widths = [x['width'] for x in root['images']]

print('Dataset Name: ',src_desc)
print('Number of images: ',n_images)
print('Number of bounding boxes: ', n_boxes)
print('Number of classes: ', n_categ)
print('Max min avg height: ', max(heights), min(heights), int(sum(heights)/len(heights)))
print('Max min avg width: ', max(widths), min(widths), int(sum(widths)/len(widths)))

[SangJunni]

Another method over there Coco dataset open for eda #5

[0seob]

sry...