使用opencv dnn readNetFromModelOptimizer

如何解决使用opencv dnn readNetFromModelOptimizer

我正在尝试使用转移学习来训练对象检测模型，以便与Intel Neural Compute Stick 2（NCS2）一起使用

到目前为止的步骤。

1. 在Google COLAB上使用tensorflow 1.15在我的自定义数据集上使用传递学习火车fast_rcnn_inception_v2_coco_2018_01_28模型。
2. 已验证的保存的tensorflow模型可与opencv-python一起使用tensorflow.saved_model.load进行对象检测
3. 冻结模型，并使用如下所示的openvino模型优化器命令创建IR .bin和.xml以与opencv-python dnn函数配合使用。

python mo_tf.py --input_model frozen_inference_graph.pb --tensorflow_object_detection_api_pipeline_config pipeline.config  --transformations_config extensions/front/tf/faster_rcnn_support_api_v1.15.json --reverse_input_channels --data_type FP16 --input_shape [1,600,3] --input image_tensor --output=detection_scores,detection_boxes,num_detections

输出如下

Model Optimizer arguments:
Common parameters:
- Path to the Input Model:  frozen_inference_graph.pb
- Path for generated IR:    /.
- IR output name:   frozen_inference_graph
- Log level:    ERROR
- Batch:    Not specified,inherited from the model
- Input layers:     image_tensor
- Output layers:    detection_scores,num_detections
- Input shapes:     [1,3]
- Mean values:  Not specified
- Scale values:     Not specified
- Scale factor:     Not specified
- Precision of IR:  FP16
- Enable fusing:    True
- Enable grouped convolutions fusing:   True
- Move mean values to preprocess section:   False
- Reverse input channels:   True

TensorFlow specific parameters:
- Input model in text protobuf format:  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:  pipeline.config
- Use the config file:  None

Model Optimizer version:    
[ WARNING ] Model Optimizer removes pre-processing block of the model which resizes image
keeping aspect ratio. The Inference Engine does not support dynamic image size so the
Intermediate Representation file is generated with the input image size of a fixed size.
The Preprocessor block has been removed. Only nodes performing mean value subtraction and
scaling (if applicable) are kept.
The graph output nodes "num_detections","detection_boxes","detection_classes","detection_scores" have been replaced with a single layer of type "Detection Output".
Refer to IR catalogue in the documentation for information about this layer.

[ WARNING ]  Network has 2 inputs overall,but only 1 of them are suitable for input
channels reversing.
Suitable for input channel reversing inputs are 4-dimensional with 3 channels
All inputs: {'image_tensor': [1,3,600],'image_info': [1,3]}
Suitable inputs {'image_tensor': [1,600]}

[ SUCCESS ] Generated IR version 10 model.
[ SUCCESS ] XML file: /./frozen_inference_graph.xml
[ SUCCESS ] BIN file: /./frozen_inference_graph.bin
[ SUCCESS ] Total execution time: 26.84 seconds. 
[ SUCCESS ] Memory consumed: 617 MB. 

1. 使用opencv-python dnn加载转换后的模型 使用openvino ubuntu_dev docker映像openvino / ubuntu18_dev：latest 我运行一个包含以下内容的python脚本。

net = cv2.dnn.readNetFromModelOptimizer('frozen_inference_graph.xml','frozen_inference_graph.bin') 
blob = cv2.dnn.blobFromImage(image_from_file)
net.setInput(blob)

报告以下错误

Traceback (most recent call last):
  File "xxxxxxxxxxxxxx-dnn.py",line 49,in <module>
  
net.setInput(blob)
cv2.error: OpenCV(4.4.0-openvino) ../opencv/modules/dnn/src/dnn.cpp:4017: error:
    (-2:Unspecified error) in function 'void   cv::dnn::dnn4_v20200609::Net::setInput(cv::InputArray,const String&,double,const Scalar&)'
    (expected: 'inputShapeLimitation.size() == blobShape.size()'),where 'inputShapeLimitation.size()' is 2 must be equal to 'blobShape.size()' is 4

请问有人可以解决此错误吗？

解决方法

我建议您尝试将模型加载到Openvino的示例中，如下所示：https://docs.openvinotoolkit.org/2018_R5/_samples_object_detection_demo_README.html

似乎使用了与blob大小有关的不兼容大小。您的python脚本可能未与动态整形相关联。

这可能对您有用：https://www.youtube.com/watch?v=Ga8j0lgi-OQ