需求:API部署内部需要将传入的base64解码后再做预测。
如果你是用tf的话,图片处理用原生tf函数要比cv2或者其它编解码库处理快的多得多的多得多!
(for TensorFlow 2)
将传入对web_safe的base64码转换为图片张量,整数类型。
# 将传入对web—safe的base64码转换为图片张量,整数类型。
def tf_decode_base64(b64_str):
imgstr_tensor = tf.convert_to_tensor(b64_str, dtype=tf.string)
tf_str = tf.io.decode_base64(imgstr_tensor, name=None)
image_tensor = tf.io.decode_image(tf_str, channels=3, dtype=tf.dtypes.uint8)
return image_tensor
测试flaskAPI常常需要用到
API调用测试:
import requests
import base64
from PIL import Image
import numpy as np
import cv2
import tensorflow as tf
# 读入图片并转为base64码,注意urlsafe_b64encode
with open('./test00.jpg', "rb") as image_file:
# encoded_image = base64.b64encode(image_file.read())
encoded_image = base64.urlsafe_b64encode(image_file.read())
req_file = encoded_image.decode('utf-8')
请求API:
url = 'http://10.1.195.41:6223/detect'
params = {"IMG_BASE64":req_file}
import time
t0=time.time()
r = requests.post(url, json=params)
print(time.time()-t0)