python-3.x - 如何用图像测试我的 CNN 模型?
问题描述
介绍/设置
我是编程新手,我根据教程制作了我的第一个 CNN 模型。我已经在 C:\Users\labadmin 中设置了我的 jupyter/tensorflow/keras
我所理解的是,我只需要从实验室管理员那里输入路径即可实施我的数据以进行测试和培训。
由于我不确定是什么导致了错误,我粘贴了整个代码和错误,我认为这是关于系统没有获取数据。
具有数据设置的文件夹如下:
labadmin 有一个名为data的文件夹,其中有两个文件夹 training和test
猫图像和狗图像都在两个文件夹中随机播放。每个文件夹有10000张图片,所以应该够了:
教程教。1. 如何创建模型 2. 定义标签 3. 创建训练数据 4. 创建和构建层 5. 创建测试数据 6. (据我了解)我创建的代码的最后一部分是
验证我的模型。
这是代码
import cv2
import numpy as np
import os
from random import shuffle
from tqdm import tqdm
TRAIN_DIR = "data\\training"
TEST_DIR = "data\\test"
IMG_SIZE = 50
LR = 1e-3
MODEL_NAME = 'dogvscats-{}-{}.model'.format(LR, '2cov-basic1')
def label_img(img):
word_label = img.split('.')[-3]
if word_label == 'cat': return [1,0]
elif word_label == 'dog': return [0,1]
def creat_train_data():
training_data = []
for img in tqdm(os.listdir(TRAIN_DIR)):
label = label_img(img)
path = os.path.join(TRAIN_DIR,img)
img = cv2.resize(cv2.imread(path, cv2.IMREAD_GRAYSCALE), (IMG_SIZE,IMG_SIZE))
training_data.append([np.array(img), np.array(label)])
shuffle(training_data)
np.save('training.npy', training_data) #save file
return training_data
import tflearn
from tflearn.layers.conv import conv_2d, max_pool_2d
from tflearn.layers.core import input_data, dropout, fully_connected
from tflearn.layers.estimator import regression
# Building convolutional convnet
convnet = input_data(shape=[None, IMG_SIZE, IMG_SIZE, 1], name='input')
# http://tflearn.org/layers/conv/
# http://tflearn.org/activations/
convnet = conv_2d(convnet, 32, 2, activation='relu')
convnet = max_pool_2d(convnet, 2)
convnet = conv_2d(convnet, 64, 2, activation='relu')
convnet = max_pool_2d(convnet, 2)
convnet = fully_connected(convnet, 1024, activation='relu')
convnet = dropout(convnet, 0.8)
#OUTPUT layer
convnet = fully_connected(convnet, 2, activation='softmax')
convnet = regression(convnet, optimizer='adam', learning_rate=LR, loss='categorical_crossentropy', name='targets')
model = tflearn.DNN(convnet, tensorboard_dir='log')
def process_test_data():
testing_data = []
for img in tqdm(os.listdir(TEST_DIR)):
path = os.path.join(TEST_DIR,img)
img_num = img.split ('.')[0] #ID of pic=img_num
img = cv2.resize(cv2-imread(path, cv2.IMREAD_GRAYSCALE), (IMG_SIZE,IMG_SIZE))
testing_data.append([np.array(img), img_num])
np.save('test_data.npy', testing_data)
return testing_data
train_data = creat_train_data()
#if you already have train data:
#train_data = np.load('train_data.npy')
100%|███████████████████████████████████████████████████████████████████████████| 21756/21756 [02:39<00:00, 136.07it/s]
if os.path.exists('{}<.meta'.format(MODEL_NAME)):
model.load(MODEL_NAME)
print('model loaded!')
train = train_data[:-500]
test = train_data[:-500]
X = np.array([i[0] for i in train]).reshape( -1, IMG_SIZE, IMG_SIZE, 1) #feature set
Y= [i[1] for i in test] #label
test_x = np.array([i[0] for i in train]).reshape( -1, IMG_SIZE, IMG_SIZE, 1)
test_y= [i[1] for i in test]
model.fit({'input': X}, {'targets': Y}, n_epoch=5, validation_set=({'input': test_x}, {'targets': test_y}),
snapshot_step=500, show_metric=True, run_id=MODEL_NAME)
Training Step: 1664 | total loss: 9.55887 | time: 63.467s
| Adam | epoch: 005 | loss: 9.55887 - acc: 0.5849 -- iter: 21248/21256
Training Step: 1665 | total loss: 9.71830 | time: 74.722s
| Adam | epoch: 005 | loss: 9.71830 - acc: 0.5779 | val_loss: 9.81653 - val_acc: 0.5737 -- iter: 21256/21256
--
三个问题
我有三个我试图解决的问题,但我没有找到解决方案:
第一个出现在:# Building convolutional convnet
curses is not supported on this machine (please install/reinstall curses for an optimal experience)
WARNING:tensorflow:From C:\Users\labadmin\Miniconda3\envs\tensorflow\lib\site-packages\tflearn\initializations.py:119: UniformUnitScaling.__init__ (from tensorflow.python.ops.init_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.initializers.variance_scaling instead with distribution=uniform to get equivalent behavior.
WARNING:tensorflow:From C:\Users\labadmin\Miniconda3\envs\tensorflow\lib\site-packages\tflearn\objectives.py:66: calling reduce_sum (from tensorflow.python.ops.math_ops) with keep_dims is deprecated and will be removed in a future version.
Instructions for updating:
keep_dims is deprecated, use keepdims instead
第二个出现在: print('model loaded!')
if os.path.exists('{}<.meta'.format(MODEL_NAME)):
model.load(MODEL_NAME)
print('model loaded!')
代码不打印的地方,是否意味着数据没有加载?
第三
教程没有介绍如何使用图像测试我的模型。那么,我可以如何以及如何添加到采用模型(也正在保存)的代码中,并从我的文件夹中运行一个图像,给定的输出是分类?
解决方案
第一:警告信息清晰,跟着它,警告就会消失。但别担心,如果你不这样做,你仍然可以正常运行你的代码。
第二:是的。如果model load!
没有打印出来,模型没有加载,检查你的模型文件的路径。
第三:要在训练后保存模型,请使用model.save("PATH-TO-SAVE")
. 然后你可以加载它model.load("PATH-TO-MODEL")
。
对于预测,使用model.predict({'input': X})
. 见这里http://tflearn.org/getting_started/#trainer-evaluator-predictor
第二个问题
- 要保存和加载模型,请使用
# Save a model
model.save('path-to-folder-you-want-to-save/my_model.tflearn')
# Load a model
model.load('the-folder-where-your-model-located/my_model.tflearn')
请记住,您应该具有模型文件的扩展名,即.tflearn
.
- 要进行预测,您需要像加载图像进行训练一样加载图像。
test_image = cv2.resize(cv2.imread("path-of-the-image", cv2.IMREAD_GRAYSCALE), (IMG_SIZE,IMG_SIZE))
test_image = np.array(test_image).reshape( -1, IMG_SIZE, IMG_SIZE, 1)
prediction = model.predict({'input': test_image })
推荐阅读
- jinja2 - Salt 命令 archive.cmd_unzip 失败,模板类型为 jinja
- javascript - Vuetify 对话框未在已安装的事件发出 vue 时打开
- python-3.x - 无法加载在 Gensim-pickle 相关错误中训练的模型
- r - 采用响应变量和预测变量并生成带有简单回归线的散点图的函数
- flutter - 无法创建 moor_database.g.dart 文件
- c - 使用 OpenCL 对大型数据集进行增量并行求和
- c# - How to get an array variable out of protected function and transfer it to another .cs file in C#
- angular - Push elements to BehavioSubject of array. Typescript Angular
- python - Python - unable to read a large file
- c# - Tasks being run synchronous