Tensorflow 进行图像分类

最新推荐文章于 2024-08-13 14:39:04 发布

简单点1024

最新推荐文章于 2024-08-13 14:39:04 发布

阅读量751

点赞数

分类专栏： DL ML

本文链接：https://blog.csdn.net/zhangbaoanhadoop/article/details/82056333

版权

本文详细介绍了如何利用Tensorflow框架进行图像分类任务，包括数据预处理、模型构建、训练过程以及模型评估。通过实例解析了卷积神经网络（CNN）在图像识别中的应用，帮助读者掌握深度学习在图像处理领域的基本操作。

摘要由CSDN通过智能技术生成

import tensorflow as tf
import loadmnist as lm
import numpy as np

def getW(shape,name):
    initial = tf.truncated_normal(shape, stddev=0.1)
    return tf.Variable(initial,name=name)
def getB(num,name):
    initial = tf.constant(0.1, shape=[num])
    return tf.Variable(initial,name=name)

def get_mean_var(ema,batch_mean,batch_var):
    ema_apply_op = ema.apply([batch_mean, batch_var])
    with tf.control_dependencies([ema_apply_op]):
        return tf.identity(batch_mean), tf.identity(batch_var)

def bn_layer(input,num):
    ema = tf.train.ExponentialMovingAverage(decay=0.5)
    batch_mean, batch_var = tf.nn.moments(input, [0, 1, 2], keep_dims=True)
    mean, var = tf.cond(is_training, lambda:(get_mean_var(ema,batch_mean,batch_var)), lambda: (ema.average(batch_mean), ema.average(batch_var)))
    shift = tf.Variable(tf.zeros([num]))
    scale = tf.Variable(tf.ones([num]))
    epsilon = 1e-3
    return tf