這是一個小的課程作業,網上大多實現的數據集是歐美人面部數據集,這裏採用的是AFAD亞洲人臉數據集,數據是由人人網採集而來,15-70+年齡段分佈。數據集圖像大小不一致,這裏統一resize到48*48大小。
數據集下載地址:http://afad-dataset.github.io/
首先resize圖像大小,這裏只處理15-56歲的圖形
# 批量處理圖片像素
from glob import glob
from PIL import Image
import os
for old in range(15, 56):
print(str(old), "年齡段批處理")
# 原始男圖片路徑
img_path_male = glob("D://dataset/AFAD-Full/AFAD-Full/" + str(old) + "/111/*.jpg")
# 原始女圖片路徑
img_path_female = glob("D://dataset/AFAD-Full/AFAD-Full/" + str(old) + "/112/*.jpg")
# 男圖保存路徑
path_save_male = "D://dataset/AFAD-Full/AFAD-Full-after-process/" + str(old) + "/0/"
# 女圖保存路徑
path_save_female = "D://dataset/AFAD-Full/AFAD-Full-after-process/" + str(old) + "/1/"
# resize男圖
a = range(0, len(img_path_male))
for i, file in enumerate(img_path_male):
name = os.path.join(path_save_male, str(old) + "-0-%d.jpg" % a[i])
im = Image.open(file)
# im.thumbnail((48, 48))
im = im.resize((48, 48))
print(im.format, im.size, im.mode)
im.save(name, 'JPEG')
# resize女圖
a = range(0, len(img_path_female))
for i, file in enumerate(img_path_female):
name = os.path.join(path_save_female, str(old) + "-1-%d.jpg" % a[i])
im = Image.open(file)
# im.thumbnail((48, 48))
im = im.resize((48, 48))
print(im.format, im.size, im.mode)
im.save(name, 'JPEG')