tensorflow-ssd 实现纸张缺陷检测

环境：win10+tensorflow1.10+python3.6.9

下载https://github.com/balancap/SSD-Tensorflow到本地

1.  解压并测试demo、

  打开Anaconda prompt ，切换到SSD-Tensorflow路径下，输入jupyter notebook

  找到ssd_notebook.ipynb并打开，修改路径并运行测试例子，相应的环境可以直接在Anaconda里安装

2. 准备纸张缺陷数据集

  拍几张纸张缺陷的图片，然后用数据增强的方式增加图片数量

  用labelImg进行特征标注，Annotions:存放图片信息的xml；JPEGImages:存放着图片

3. 修改datasets文件夹中pascalvoc_common.py文件，将训练类修改别成自己的。

4. 运行tf_convert_data.py文件 

python tf_convert_data.py -data_name=pascalvoc -dataset_dir=D:\tensorflow-ssd\data -output_ name=voc_2007_test -output_dir=D:\tensorflow-ssd\tfrecords

5. 训练模型train_ssd_network.py文件中修改

6. 运行train_ssd_network.py

python  train_ssd_network.py  -train_dir=D:\tensorflow-ssd\SSD-Tensorflow-master\logs  -dataset_dir=D:\tensorflow-ssd\tfrecords  -dataset_name=pascalvoc_2007  -dataset_split_name=test  -model_name=ssd_300_vgg   -checkpoint_path=D:\tensorflow-ssd\SSD-Tensorflow-master\checkpoints\vgg_16.ckpt   -checkpoint_model_scope=vgg_16  -checkpoint_exclude_scopes=ssd_300_vgg/conv6,ssd_300_vgg/conv7,ssd_300_vgg/block8,ssd_300_vgg/block9,ssd_300_vgg/block10,ssd_300_vgg/block11,ssd_300_vgg/block4_box,ssd_300_vgg/block7_box,ssd_300_vgg/block8_box,ssd_300_vgg/block9_box,ssd_300_vgg/block10_box,ssd_300_vgg/block11_box   -trainable_scopes=ssd_300_vgg/conv6,ssd_300_vgg/conv7,ssd_300_vgg/block8,ssd_300_vgg/block9,ssd_300_vgg/block10,ssd_300_vgg/block11,ssd_300_vgg/block4_box,ssd_300_vgg/block7_box,ssd_300_vgg/block8_box,ssd_300_vgg/block9_box,ssd_300_vgg/block10_box,ssd_300_vgg/block11_box  -save_summaries_secs=60   -save_interval_secs=600 --weight_decay=0.0005    -optimizer=adam    -learning_rate=0.001  -learning_rate_decay_factor=0.94    -batch_size=4

python  train_ssd_network.py  -train_dir=D:\tensorflow-ssd\SSD-Tensorflow-master\logs  -dataset_dir=D:\tensorflow-ssd\tfrecords  -dataset_name=pascalvoc_2007  -dataset_split_name=test  -model_name=ssd_300_vgg   -checkpoint_path=D:\tensorflow-ssd\SSD-Tensorflow-master\checkpoints\vgg_16.ckpt   -checkpoint_model_scope=vgg_16  -checkpoint_exclude_scopes=ssd_300_vgg/conv6,ssd_300_vgg/conv7,ssd_300_vgg/block8,ssd_300_vgg/block9,ssd_300_vgg/block10,ssd_300_vgg/block11,ssd_300_vgg/block4_box,ssd_300_vgg/block7_box,ssd_300_vgg/block8_box,ssd_300_vgg/block9_box,ssd_300_vgg/block10_box,ssd_300_vgg/block11_box   -trainable_scopes=ssd_300_vgg/conv6,ssd_300_vgg/conv7,ssd_300_vgg/block8,ssd_300_vgg/block9,ssd_300_vgg/block10,ssd_300_vgg/block11,ssd_300_vgg/block4_box,ssd_300_vgg/block7_box,ssd_300_vgg/block8_box,ssd_300_vgg/block9_box,ssd_300_vgg/block10_box,ssd_300_vgg/block11_box  -save_summaries_secs=60   -save_interval_secs=600 --weight_decay=0.0005    -optimizer=adam    -learning_rate=0.001  -learning_rate_decay_factor=0.94    -batch_size=4

7. 训练完成后，在logs文件夹中会有 model.ckpt-10000.data-00000-of-00001 and model.ckpt-10000.index。把这两个文件复制到checkpoints文件夹，修改名字为model.ckpt.data-00000-of-00001 and model.ckpt.index

8. 测试，打开之前的demo，修改ckpt_filename='D:\tensorflow-ssd\SSD-Tensorflow-master\checkpoints/model.ckpt'，然后在demo文件夹里放入测试图片，运行即可。