openvino部署目标检测

安装须知

• 确保安装成功OpenVINO，最好是2017R3之后的版本。本文使用的版本为2018R4
• 确保安装tensorflow，本文的tensorflow为源码编译的tensorflow-gpu-1.10(编译最新版的tensorflow出错)
• ubuntu16.0.4 目前为止OpenVINO不支持ubuntu18.0.4

下载ssdv2版本的目标检测训练压缩文件

本文下载的ssd模型地址为下载链接，其它地方下载的模型并没有测试过，在后续将测试Google Model Zone的模型是否可行。
生成xml中间文件

这里需要注意的是，Intel官方教程给的是./mo_tf.py这里应该是不对的（给mo_tf.py添加执行权限后也许可以，没有尝试，另外直接python3 mo_tf.py --input_model ssd_inception_v2_coco_2018_01_28/frozen_inference_graph.pb会报错，加上input_shape也会报错目前只有使用json+pipeline.config不报错）

python3 mo_tf.py --input_model ssd_inception_v2_coco_2018_01_28/frozen_inference_graph.pb --output=detection_boxes,detection_scores,num_detections --tensorflow_use_custom_operations_config /home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/extensions/front/tf/ssd_v2_support.json --tensorflow_object_detection_api_pipeline_config /home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/ssd_inception_v2_coco_2018_01_28/pipeline.config

输出如下：

Model Optimizer arguments:
Common parameters:
	- Path to the Input Model: 	/home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/ssd_inception_v2_coco_2018_01_28/frozen_inference_graph.pb
	- Path for generated IR: 	/home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/.
	- IR output name: 	frozen_inference_graph
	- Log level: 	ERROR
	- Batch: 	Not specified, inherited from the model
	- Input layers: 	Not specified, inherited from the model
	- Output layers: 	detection_boxes,detection_scores,num_detections
	- Input shapes: 	Not specified, inherited from the model
	- Mean values: 	Not specified
	- Scale values: 	Not specified
	- Scale factor: 	Not specified
	- Precision of IR: 	FP32
	- Enable fusing: 	True
	- Enable grouped convolutions fusing: 	True
	- Move mean values to preprocess section: 	False
	- Reverse input channels: 	False
TensorFlow specific parameters:
	- Input model in text protobuf format: 	False
	- Offload unsupported operations: 	False
	- Path to model dump for TensorBoard: 	None
	- List of shared libraries with TensorFlow custom layers implementation: 	None
	- Update the configuration file with input/output node names: 	None
	- Use configuration file used to generate the model with Object Detection API: 	/home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/ssd_inception_v2_coco_2018_01_28/pipeline.config
	- Operations to offload: 	None
	- Patterns to offload: 	None
	- Use the config file: 	/home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/extensions/front/tf/ssd_v2_support.json
Model Optimizer version: 	1.4.292.6ef7232d
/home/amax/anaconda3/lib/python3.5/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
  from ._conv import register_converters as _register_converters
The Preprocessor block has been removed. Only nodes performing mean value subtraction and scaling (if applicable) are kept.

[ SUCCESS ] Generated IR model.
[ SUCCESS ] XML file: /home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/./frozen_inference_graph.xml
[ SUCCESS ] BIN file: /home/amax/intel/computer_vision_sdk_2018.4.420/deployment_tools/model_optimizer/./frozen_inference_graph.bin
[ SUCCESS ] Total execution time: 20.20 seconds. 

在当前目录下将生成文件：

• frozen_inference_graph.xml
• frozen_inference_graph.bin
• frozen_inference_graph.mapping
  生成的下载文件可以在这里下载：
  

使用中间文件推理

运行脚本如下：

注意：输入图像可以是jpeg格式但是输出为bmp格式。

ssd_bin=/home/amax/inference_engine_samples_build/intel64/Release/object_detection_sample_ssd
network=/home/amax/intel/computer_vision_sdk/deployment_tools/ssd_detect/frozen_inference_graph.xml
${ssd_bin} -i example.bmp -m ${network} -d CPU 

输出结果如下：

[ INFO ] InferenceEngine:
        API version ............ 1.4
        Build .................. 17328
Parsing input parameters
[ INFO ] Files were added: 1
[ INFO ]     example.bmp
[ INFO ] Loading plugin

        API version ............ 1.4
        Build .................. lnx_20181004
        Description ....... MKLDNNPlugin
[ INFO ] Loading network files:
        /home/amax/intel/computer_vision_sdk/deployment_tools/ssd_detect/frozen_inference_graph.xml
        /home/amax/intel/computer_vision_sdk/deployment_tools/ssd_detect/frozen_inference_graph.bin
[ INFO ] Preparing input blobs
[ INFO ] Batch size is 1
[ INFO ] Preparing output blobs
[ INFO ] Loading model to the plugin
[ WARNING ] Image is resized from (640, 747) to (300, 300)
[ INFO ] Batch size is 1
[ INFO ] Start inference (1 iterations)
[ INFO ] Processing output blobs
[0,1] element, prob = 0.912171    (28.3366,4.0617)-(640,743.86) batch id : 0 WILL BE PRINTED!
[ INFO ] Image out_0.bmp created!

total inference time: 28.7588
Average running time of one iteration: 28.7588 ms

Throughput: 34.772 FPS

[ INFO ] Execution successful

原图：

输出结果

model zone测试通过

• Faster RCNN_Intepction_v2