flickr-img-only-imgcap-dict

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from pickle import dump

#Standard variable
IMAGES_PATH = "data/Flicker8k_Dataset/"
TOKENS_PATH = "data/Flickr8k_text/"

def image_to_caption_dict(image_path=IMAGES_PATH, tokens_path=TOKENS_PATH):
    with open(tokens_path+"Flickr8k.token.txt") as f:
        data = f.read()

    descriptions = dict()

    for el in data.strip().split("\n"):

        tokens = el.split()
        image_id , image_desc = tokens[0],tokens[1:]

        # dropping .jpg from image id
        image_id = image_id.split(".")[0]

        image_desc = " ".join(image_desc)

        if image_id in descriptions:
            descriptions[image_id].append(image_desc)
        else:
            descriptions[image_id] = [image_desc]

    return descriptions

descriptions = image_to_caption_dict()


descriptions["1000268201_693b08cb0e"]

tmp = list(descriptions.keys())[8090]
from IPython.display import Image, display
z = Image(filename=IMAGES_PATH+tmp+'.jpg')
display(z)

for cap in descriptions[tmp]:
    print(cap)