1.ION：在conv3、conv4、conv5和contextfeatures上分别进行roi_pooling,在channel那一维进行concat2.Hypernet:在较浅层max_pooling，中间层保持不变，较高层deconv，最后把这三个结果concat起来构成最后一层featuremapION是在不同...

#1.自定义模型并训练。importnumpyasnpimporttensorflowastffromtensorflow.examples.tutorials.mnistimportinput_datatf.logging.set_verbosity(tf.logging.INFO)deflenet(x,is_tra...

#1.模型定义。importnumpyasnpimporttensorflowastffromtensorflow.examples.tutorials.mnistimportinput_datatf.logging.set_verbosity(tf.logging.INFO)mnist=input_data.read...

#1.模型定义。importtensorflowastffromtensorflow.examples.tutorials.mnistimportinput_datamnist_data=input_data.read_data_sets('F:\TensorFlowGoogle\201806-github\datas...

#1.数据预处理。importkerasfromkeras.modelsimportModelfromkeras.datasetsimportmnistfromkeras.layersimportInput,Densefromtflearn.layers.coreimportfully_connectednum_cla...

#1.数据预处理。importkerasfromkeras.modelsimportModelfromkeras.datasetsimportmnistfromkeras.layersimportInput,Densefromtflearn.layers.coreimportfully_connectednum_cla...

#1.数据预处理。fromkeras.layersimportLSTMfromkeras.datasetsimportimdbfromkeras.modelsimportSequentialfromkeras.preprocessingimportsequencefromkeras.layersimportDense,...

#1.数据预处理importkerasfromkerasimportbackendasKfromkeras.datasetsimportmnistfromkeras.modelsimportSequentialfromkeras.layersimportDense,Flatten,Conv2D,MaxPooling2D...

  将原来版本的keras卸载了，再安装2.1.5版本的keras就可以了。...

找到对应的keras配置文件keras.json  将里面的内容修改为以下就可以了 ...

#1.通过TFLearn的API定义卷机神经网络。importtflearnimporttflearn.datasets.mnistasmnistfromtflearn.layers.convimportconv_2d,max_pool_2dfromtflearn.layers.estimatorimportregre...

#1.通过TensorFlow-Slim定义卷机神经网络importnumpyasnpimporttensorflowastfimporttensorflow.contrib.slimasslimfromtensorflow.examples.tutorials.mnistimportinput_data#通过Tens...

importtempfileimporttensorflowastf#1.列举输入文件。#输入数据生成的训练和测试数据。train_files=tf.train.match_filenames_once("F:\output.tfrecords")test_files=tf.train.match_filenames_...