操作系统:windows 10  64位

内存:8G

GPU:Nvidia MX 150

Tensorflow: 1.4

1、安装python

Anaconda3-5.0.1 ,默认python版本(3.6.3)

2、安装tensorflow

pip install --upgrade tensorflow

conda install pip #更新pip

pip install --upgrade tensorflow-gpu

CUDA® Toolkit 8.0, 需要注意最新版9.1不支持tensorflow 1.4版本;

cuDNN v6.0,7.0不支持tensorflow 1.4版本,现在cuDNN需要先注册成为NVIDIA开发者,下载后将cuDNN中对应文件夹下的.dll文件分别复制到CUDA安装目录对应文件夹下;

对应的显卡驱动,如果驱动较新,在安装CUDA的时候会有提示可能不兼容,可以无视。

3、测试gpu版安装好了没有

improt tensorflow as tf
hello = tf.constant('hello')
sess = tf.Session()
print(sess.run(hello)) 当输出hello则装好tensorflow from tensorflow.python.client import device_lib
print(device_lib.list_local_devices()) 当输出:
Sample Output
[name: "/cpu:0" device_type: "CPU" memory_limit:
name: "/gpu:0" device_type: "GPU" .............GeForce GTX 1070

4、下载api

https://github.com/tensorflow/models

5、protobuf配置

https://github.com/google/protobuf/releases  网站中选择windows 版本(最下面),解压后将bin文件夹中的【protoc.exe】放到C:\Windows

在models\research\目录下打开命令行窗口,输入:

# From tensorflow/models/
protoc object_detection/protos/*.proto --python_out=.

在这一步有时候会出错,可以尝试把/*.proto 这部分改成文件夹下具体的文件名,一个一个试,每运行一个,文件夹下应该出现对应的.py结尾的文件。不报错即可。

6、环境变量

models/research/ 及 models/research/slim 添加进环境变量

7、测试环境

python object_detection/builders/model_builder_test.py

注意 :如果出现no model name object_api这个东东,就在D:\anaconda\anaconda3.4.2.0\Lib\site-packages目录下面新建一个my_objection.pth文件,文件内容就是这两个路径,如下图:

8、object_detection_tutorial.ipynb

代码简化了一些。

# coding: utf-8
import numpy as np
import os
import six.moves.urllib as urllib
import sys
import tarfile
import tensorflow as tf
import zipfile
from collections import defaultdict
from io import StringIO
from matplotlib import pyplot as plt
from PIL import Image if tf.__version__ < '1.4.0':
raise ImportError('Please upgrade your tensorflow installation to v1.4.* or later!') # This is needed to display the images.
get_ipython().magic('matplotlib inline') # This is needed since the notebook is stored in the object_detection folder.
sys.path.append("..") from utils import label_map_util
from utils import visualization_utils as vis_util # 下载模型名,设置对应的参数
MODEL_NAME = 'ssd_mobilenet_v1_coco_2017_11_17'
MODEL_FILE = MODEL_NAME + '.tar.gz'
DOWNLOAD_BASE = 'http://download.tensorflow.org/models/object_detection/' # 训练好的模型,用来检测
PATH_TO_CKPT = MODEL_NAME + '/frozen_inference_graph.pb' # 标签文件,记录了哪些标签需要识别
PATH_TO_LABELS = os.path.join('data', 'mscoco_label_map.pbtxt') # 类别数目,根据实际修改
NUM_CLASSES = 90 # ## 下载上面说的模型(不用改)
opener = urllib.request.URLopener()
opener.retrieve(DOWNLOAD_BASE + MODEL_FILE, MODEL_FILE)
tar_file = tarfile.open(MODEL_FILE)
for file in tar_file.getmembers():
file_name = os.path.basename(file.name)
if 'frozen_inference_graph.pb' in file_name:
tar_file.extract(file, os.getcwd()) #将训练完的载入内存(不用改)
detection_graph = tf.Graph()
with detection_graph.as_default():
od_graph_def = tf.GraphDef()
with tf.gfile.GFile(PATH_TO_CKPT, 'rb') as fid:
serialized_graph = fid.read()
od_graph_def.ParseFromString(serialized_graph)
tf.import_graph_def(od_graph_def, name='') # ## 载入标签map(不用改)
# Label maps map indices to category names, so that when our convolution network predicts `5`,
we know that this corresponds to `airplane`.
Here we use internal utility functions, but anything that returns a dictionary mapping integers to appropriate string labels would be fine label_map = label_map_util.load_labelmap(PATH_TO_LABELS)
categories = label_map_util.convert_label_map_to_categories(label_map, max_num_classes=NUM_CLASSES, use_display_name=True)
category_index = label_map_util.create_category_index(categories) def load_image_into_numpy_array(image):
(im_width, im_height) = image.size
return np.array(image.getdata()).reshape(
(im_height, im_width, 3)).astype(np.uint8) """
检测部分
"""
# 目标文件夹名
PATH_TO_TEST_IMAGES_DIR = 'test_images'
# 源码中test_images文件夹下就两张image,名字分别为image1.jpg和image2.jpg
# 如果想用自己的image,有5张图片,分别为hello1.jpg.....hello5.jpg可以改成:
# TEST_IMAGE_PATHS = [ os.path.join(PATH_TO_TEST_IMAGES_DIR, 'hello{}.jpg'.format(i)) for i in range(1, 6) ]
TEST_IMAGE_PATHS = [ os.path.join(PATH_TO_TEST_IMAGES_DIR, 'image{}.jpg'.format(i)) for i in range(1, 3) ] # 设置输出图像的英尺
IMAGE_SIZE = (12, 8) #运行,进行检测
with detection_graph.as_default():
with tf.Session(graph=detection_graph) as sess:
# Definite input and output Tensors for detection_graph
image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
# Each box represents a part of the image where a particular object was detected.
detection_boxes = detection_graph.get_tensor_by_name('detection_boxes:0')
# Each score represent how level of confidence for each of the objects.
# Score is shown on the result image, together with the class label.
detection_scores = detection_graph.get_tensor_by_name('detection_scores:0')
detection_classes = detection_graph.get_tensor_by_name('detection_classes:0')
num_detections = detection_graph.get_tensor_by_name('num_detections:0')
for image_path in TEST_IMAGE_PATHS:
image = Image.open(image_path)
# the array based representation of the image will be used later in order to prepare the
# result image with boxes and labels on it.
image_np = load_image_into_numpy_array(image)
# Expand dimensions since the model expects images to have shape: [1, None, None, 3]
image_np_expanded = np.expand_dims(image_np, axis=0)
# Actual detection.
(boxes, scores, classes, num) = sess.run(
[detection_boxes, detection_scores, detection_classes, num_detections],
feed_dict={image_tensor: image_np_expanded})
# Visualization of the results of a detection.
vis_util.visualize_boxes_and_labels_on_image_array(
image_np,
np.squeeze(boxes),
np.squeeze(classes).astype(np.int32),
np.squeeze(scores),
category_index,
use_normalized_coordinates=True,
line_thickness=8)
plt.figure(figsize=IMAGE_SIZE)
plt.imshow(image_np)

ssd_mobilenet_demo的更多相关文章

随机推荐

  1. 解决Chrome网页编码显示乱码的问题

    解决Chrome网页编码显示乱码的问题 记得在没多久以前,Google Chrome上面出现编码显示问题时,可以手动来调整网页编码问题,可是好像在Chrome 55.0版以后就不再提供手动调整编码,所 ...

  2. CentOS 6 安装Syslog-ng

    entOS 6 安装 Syslog-ng 一. yum 安装 syslog-ng3.7.1 是专门用于RHEL/CentOS version 6 ,不要安装成其他版本.亲身经历,安装成syslog-n ...

  3. 从Android手机中取出已安装的app包,导出apk

    从Android手机中取出已安装的app包,导出apk TAG:Android,提取,apk,adb,pm,root,导出apk 假设有这样一个场景,A君看到你手机上一个实用APP,想要安装到自己手机 ...

  4. python爬虫——爬取淘票票正在热映电影

    今天正好学习了一下python的爬虫,觉得收获蛮大的,所以写一篇博客帮助想学习爬虫的伙伴们. 这里我就以一个简单地爬取淘票票正在热映电影为例,介绍一下一个爬虫的完整流程. 首先,话不多说,上干货——源 ...

  5. Queen Attack -- 微软2017年预科生计划在线编程笔试第二场

    #!/usr/bin/env python # coding:utf-8 # Queen Attack # https://hihocoder.com/problemset/problem/1497 ...

  6. type动态创建类

    在一些特定场合,需要动态创建类,比如创建表单,就会用到type动态创建类,举个例子: class Person(object): def __init__(self,name,age): self.n ...

  7. SQL复制远程数据库数据到本地-及查询结果少显示一列

    网上找了查询结果怎么少显示一列,因为数据很多列,结果不是视图就是嵌套,太麻烦,这里用临时表做 exec sp_addlinkedserver 'ITSV ', ' ', 'SQLOLEDB', '19 ...

  8. 小白学Python(11)——pyecharts,绘制饼图 Pie

    Pie-基本示例 from example.commons import Faker from pyecharts import options as opts from pyecharts.char ...

  9. 问题 G: 圆桌上的晚餐

    问题 G: 圆桌上的晚餐 时间限制: 1 Sec  内存限制: 128 MB提交: 1583  解决: 656[提交] [状态] [命题人:jsu_admin] 题目描述         大家一定在圆 ...

  10. javascript中slice(),splice(),split(),substring(),substr()使用方法

    因为本人在使用这些方法时常有混淆,特总结如下: 1.slice(): Array和String对象都有 在Array中  slice(i,[j]) i为开始截取的索引值,负数代表从末尾算起的索引值,- ...