使用TensorFlow识别照片中的物体
Posted 熊猫滚滚
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了使用TensorFlow识别照片中的物体相关的知识,希望对你有一定的参考价值。
1.环境ubuntu14.04.5
安装TensorFlow
官方文档:https://www.tensorflow.org/install/install_linux
sudo pip install --upgrade https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.1-cp27-none-linux_x86_64.whl
注意:后面的网址根据不同设置而不同:
1 Python 2.7 2 3 CPU only: 4 5 https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.1-cp27-none-linux_x86_64.whl 6 GPU support: 7 8 https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-1.0.1-cp27-none-linux_x86_64.whl 9 Note that GPU support requires the NVIDIA hardware and software described in NVIDIA requirements to run TensorFlow with GPU support. 10 11 Python 3.4 12 13 CPU only: 14 15 https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.1-cp34-cp34m-linux_x86_64.whl 16 GPU support: 17 18 https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-1.0.1-cp34-cp34m-linux_x86_64.whl 19 Note that GPU support requires the NVIDIA hardware and software described in NVIDIA requirements to run TensorFlow with GPU support. 20 21 Python 3.5 22 23 CPU only: 24 25 https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.1-cp35-cp35m-linux_x86_64.whl 26 GPU support: 27 28 https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-1.0.1-cp35-cp35m-linux_x86_64.whl 29 Note that GPU support requires the NVIDIA hardware and software described in NVIDIA requirements to run TensorFlow with GPU support. 30 31 Python 3.6 32 33 CPU only: 34 35 https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.1-cp36-cp36m-linux_x86_64.whl 36 GPU support: 37 38 https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-1.0.1-cp36-cp36m-linux_x86_64.whl
我选择了相当于:python2.7 纯CPU
自动安装无误,若有安装失败,可能是网络问题,重新执行一遍试试。
也有可能是若干依赖没有装好:先把 各种依赖都安一遍
下载神经网络模型:http://download.tensorflow.org/models/image/imagenet/inception-2015-12-05.tgz
1 mkdir model 2 tar -xf archive_name.tar -C model/
下载代码:https://codeload.github.com/tensorflow/models/zip/master
1 unzip models-master.zip 2 cd models-master/tutorials/image/imagenet/
其中的classify_image.py是我们实现物体识别的主代码
接下来做几个测试
1 classify_image.py --model_dir /home/usr/model --image_file /home/usr/model/cropped_panda.jpg 2 classify_image.py --model_dir /home/usr/model --image_file /home/usr/model/testkb.jpg 3 classify_image.py --model_dir /home/usr/model --image_file /home/usr/model/testmous.jpg
1 giant panda, panda, panda bear, coon bear, ailuropoda melanoleuca (score = 0.89107) 2 computer keyboard, keypad (score = 0.64016) 3 mouse, computer mouse(score = 0.89453)
以上是关于使用TensorFlow识别照片中的物体的主要内容,如果未能解决你的问题,请参考以下文章
业界 | 谷歌开源TensorFlow Object Detection API物体识别系统
深度学习基于tensorflow的小型物体识别训练(数据集:CIFAR-10)
玩转谷歌物体识别API,用TensorFlow和OpenCV打造实时识别应用