resnet将图片与文本作为输入，进行二分类

#!/usr/bin/env python
# coding: utf-8

# In[1]:

import os
os.environ["CUDA_VISIBLE_DEVICES"] = '0'

# In[2]:

import tensorflow as tf

# In[3]:

keras = tf.keras
layers = tf.keras.layers

# In[4]:

# 2.0.0-alpha0

# In[5]:

import matplotlib.pyplot as plt

# In[6]:

get_ipython().run_line_magic('matplotlib', 'inline')

# In[7]:

import numpy as np

# In[8]:

import glob

# In[9]:

import os

# In[10]:

print("Tensorflow version:{}".format(tf.__version__))

# In[11]:

total_image = glob.glob("./data/*")

# In[12]:

# 过滤不能加载的图片
# 单进程
# total_image_ = []
# l = len(total_image)
# # l = 1000
# for idx in range(l):
#     try:
#         image = tf.io.read_file(total_image[idx])
#         tf.image.decode_jpeg(image, channels=3)
#         total_image_.append(total_image[idx])
#     except:
#         print(idx)
#         print(total_image[idx])

# In[13]:

# from multiprocessing import Process
# from multiprocessing import Queue
# q = Queue()

from multiprocessing import Pool
alist = []
def one_work(a, b):
    alist = []
    for idx in range(a,b):
        if not total_image[idx].endswith(".jpg"):
            continue
            
        try:
            image = tf.io.read_file(total_image[idx])
            tf.image.decode_jpeg(image, channels=3)
            alist.append(total_image[idx])
        except:
            pass
    return alist
pool = Pool(processes=16)
l = len(total_image)
one = l//32 + 1
results = []
for idx in range(32):
    results.append(pool.apply_async(one_work, (idx*one, min(idx*one+one,l))))
pool.close()
pool.join()

# In[14]:

len(results)

# In[15]:

total_image_ = []
for idx in range(len(results)):
    result = results[idx]
    total_image_ += result.get()

# In[16]:

len(total_image_)

# In