电影口碑与海报图像的相关性分析

 1 # -*- coding: utf-8 -*-
 2 
 3 import pandas as pd
 4 from tools import get_n_face, get_color_mean
 5 import matplotlib.pyplot as plt
 6 
 7 # 数据集路径
 8 dataset_path = ‘./dataset/movie_metadata.csv‘
 9 
10 # 是否第一次运行
11 is_first_run = False
12 
13 
14 def run_main():
15     """
16         主函数
17     """
18     if is_first_run:
19         # 第一次运行程序
20         data_df = pd.read_csv(dataset_path, nrows=100)
21 
22         # 记录海报中人脸个数
23         print(‘海报人脸检测：‘)
24         data_df[‘n_face‘] = data_df[‘movie_imdb_link‘].apply(get_n_face)
25 
26         # 保存结果
27         data_df.to_csv(‘./imdb_face.csv‘, index=False, encoding=‘utf-8‘)
28 
29         # 记录海报的平均像素值（这里代表图像的主要颜色，可通过颜色的表达方式替代）
30         print(‘海报像素均值计算：‘)
31         data_df[‘color_mean‘] = data_df[‘movie_imdb_link‘].apply(get_color_mean)
32 
33         # 保存结果
34         data_df.to_csv(‘./imdb_face_color.csv‘, index=False, encoding=‘utf-8‘)
35 
36     data_df = pd.read_csv(‘./imdb_face_color.csv‘)
37     # 分析结果
38     # 人脸个数和评分的关系
39     # 去除无效人脸
40     data_df = data_df[data_df[‘n_face‘] != -1]
41     face_score = data_df[‘imdb_score‘].groupby(data_df[‘n_face‘]).mean()
42     face_score.name = ‘Score‘
43     face_score.index.name = ‘Number of Face‘
44     face_score.plot(kind=‘bar‘)
45     plt.show()
46 
47     # 像素均值和评分的关系
48     color_score = data_df[‘imdb_score‘].groupby(data_df[‘color_mean‘]).mean()
49     color_score.name = ‘Score‘
50     color_score.index.name = ‘Mean of color‘
51     color_score.plot(kind=‘bar‘)
52     plt.show()
53 
54 
55 if __name__ == ‘__main__‘:
56     run_main()

 1 # -*- coding: utf-8 -*-
 2 
 3 from bs4 import BeautifulSoup
 4 import urllib.request
 5 import cognitive_face as CF
 6 from skimage import io
 7 import numpy as np
 8 
 9 
10 def get_img_link(movie_link):
11     """
12         通过电影的链接爬取海报的链接
13     """
14     movie_html = urllib.request.urlopen(movie_link)
15     movie_html_obj = BeautifulSoup(movie_html, ‘html.parser‘, from_encoding=‘utf-8‘)
16     # 获取海报小图的链接
17     small_poster_img_link = movie_html_obj.find(‘div‘, class_=‘poster‘).find(‘img‘)[‘src‘]
18 
19     # 获取海报大图的链接
20     big_poster_img_link = small_poster_img_link[:small_poster_img_link.find(‘._V1_‘) + 4] + ‘.jpg‘
21 
22     return big_poster_img_link
23 
24 
25 def get_n_face(movie_link):
26     """
27         通过图像链接获取包含的人脸个数
28     """
29     print(‘正在处理链接：‘, movie_link)
30     img_link = get_img_link(movie_link)
31     Key = ‘xxxxxxxxxx ‘  # 这里请替换成自己申请的key
32     CF.Key.set(Key)
33     n_face = -1
34     try:
35         face_list = CF.face.detect(img_link)
36         n_face = len(face_list)
37         print(‘人脸个数：‘, n_face)
38     except CF.util.CognitiveFaceException:
39         print(‘无效图片‘)
40     return n_face
41 
42 
43 def round_to_int(x, base=10):
44     """
45         将数字转换到最近的整数
46     """
47     return int(base * round(float(x)/base))
48 
49 
50 def get_color_mean(movie_link):
51     """
52         通过图像链接获取其平均像素值
53     """
54     print(‘正在处理链接：‘, movie_link)
55     img_link = get_img_link(movie_link)
56     image = io.imread(img_link)
57     int_mean_color = round_to_int(np.mean(image))
58     print(‘像素均值：‘, int_mean_color)
59     return int_mean_color