Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.
6404 Discussions

Problem Converting Tensorflow Model (Illegal Instruction)

Rivera__Maverick
Beginner
‎02-17-2020 02:08 AM
457 Views

I used the code below to convert my model. But it does not work. 

 sudo python3 mo_tf.py --input_model /home/ls201/Desktop/veh_class_faster_rcnn/frozen_inference_graph.pb --transformations_config /opt/intel/openvino/deployment_tools/model_optimizer/extensions/front/tf/faster_rcnn_support_api_v1.14.json --tensorflow_object_detection_api_pipeline_config /home/ls201/Desktop/veh_class_faster_rcnn/pipeline.config

the output is:

Model Optimizer arguments:
Common parameters:
    - Path to the Input Model:     /home/ls201/Desktop/veh_class_faster_rcnn/frozen_inference_graph.pb
    - Path for generated IR:     /opt/intel/openvino_2020.1.023/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:     Not specified, inherited from the model
    - 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
    - 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/ls201/Desktop/veh_class_faster_rcnn/pipeline.config
    - Operations to offload:     None
    - Patterns to offload:     None
    - Use the config file:     None
Model Optimizer version:     2020.1.0-61-gd349c3ba4a
Illegal instruction
 

link for model and pipeline config

https://drive.google.com/open?id=1vpVoYAfA5IdOcXNLgyelEUkENRHLRW_P

0 Kudos

Link Copied

2 Replies
JesusE_Intel
Moderator
‎02-18-2020 12:18 PM
457 Views

Hi Maverick,

It looks like the pipeline.config file is empty, could you double check?

Also, please share additional information about the model.

  • Is it a pre-trained model or custom trained model?
  • If it's custom trained model, what base model was used for training?
  • Which version of TensorFlow was used for training?
  • How did you freeze the model?

Regards,

Jesus

 

0 Kudos
Copy link
Viramontes_Olivar__J
Beginner
‎06-13-2020 04:31 PM
457 Views

I have the same problem but when I'm installing the robot_devkit, it says "Illegal instruction", this is the output:

 

Model Optimizer arguments:
Common parameters:
    - Path to the Input Model:     /home/ramon/Downloads/models/mask_rcnn_inception_v2_coco_2018_01_28/frozen_inference_graph.pb
    - Path for generated IR:     /opt/openvino_toolkit/models/segmentation/output/FP32
    - 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:     Not specified, inherited from the model
    - 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:     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:     /home/ramon/Downloads/models/mask_rcnn_inception_v2_coco_2018_01_28/pipeline.config
    - Use the config file:     /opt/intel/openvino/deployment_tools/model_optimizer/extensions/front/tf/mask_rcnn_support.json
Model Optimizer version:     
/home/ramon/robot_devkit/packages/perception/deps/33-openvino.deps: line 14: 18813 Illegal instruction     sudo python3 /opt/intel/openvino/deployment_tools/model_optimizer/mo_tf.py --input_model frozen_inference_graph.pb --tensorflow_use_custom_operations_config /opt/intel/openvino/deployment_tools/model_optimizer/extensions/front/tf/mask_rcnn_support.json --tensorflow_object_detection_api_pipeline_config pipeline.config --reverse_input_channels --output_dir /opt/openvino_toolkit/models/segmentation/output/FP32

 

Because of this it can't finish the process of installation of robot_devkit, specifically the problem comes out when I run the command: ./demo/rdk_install.sh , I'm following the instructions of this link: https://intel.github.io/robot_devkit_doc/pages/install.html

 

I've already have the two cameras and the neural compute stick 2,

 

the file content of the file: mask_rcnn_inception_v2_coco_2018_01_28/pipeline.config

 

is:

model {
  faster_rcnn {
    number_of_stages: 3
    num_classes: 90
    image_resizer {
      keep_aspect_ratio_resizer {
        min_dimension: 800
        max_dimension: 1365
      }
    }
    feature_extractor {
      type: "faster_rcnn_inception_v2"
      first_stage_features_stride: 16
    }
    first_stage_anchor_generator {
      grid_anchor_generator {
        height_stride: 16
        width_stride: 16
        scales: 0.25
        scales: 0.5
        scales: 1.0
        scales: 2.0
        aspect_ratios: 0.5
        aspect_ratios: 1.0
        aspect_ratios: 2.0
      }
    }
    first_stage_box_predictor_conv_hyperparams {
      op: CONV
      regularizer {
        l2_regularizer {
          weight: 0.0
        }
      }
      initializer {
        truncated_normal_initializer {
          stddev: 0.00999999977648
        }
      }
    }
    first_stage_nms_score_threshold: 0.0
    first_stage_nms_iou_threshold: 0.699999988079
    first_stage_max_proposals: 100
    first_stage_localization_loss_weight: 2.0
    first_stage_objectness_loss_weight: 1.0
    initial_crop_size: 14
    maxpool_kernel_size: 2
    maxpool_stride: 2
    second_stage_box_predictor {
      mask_rcnn_box_predictor {
        fc_hyperparams {
          op: FC
          regularizer {
            l2_regularizer {
              weight: 0.0
            }
          }
          initializer {
            variance_scaling_initializer {
              factor: 1.0
              uniform: true
              mode: FAN_AVG
            }
          }
        }
        use_dropout: false
        dropout_keep_probability: 1.0
        conv_hyperparams {
          op: CONV
          regularizer {
            l2_regularizer {
              weight: 0.0
            }
          }
          initializer {
            truncated_normal_initializer {
              stddev: 0.00999999977648
            }
          }
        }
        predict_instance_masks: true
        mask_prediction_conv_depth: 0
        mask_height: 15
        mask_width: 15
        mask_prediction_num_conv_layers: 2
      }
    }
    second_stage_post_processing {
      batch_non_max_suppression {
        score_threshold: 0.300000011921
        iou_threshold: 0.600000023842
        max_detections_per_class: 100
        max_total_detections: 100
      }
      score_converter: SOFTMAX
    }
    second_stage_localization_loss_weight: 2.0
    second_stage_classification_loss_weight: 1.0
    second_stage_mask_prediction_loss_weight: 4.0
  }
}
train_config {
  batch_size: 1
  data_augmentation_options {
    random_horizontal_flip {
    }
  }
  optimizer {
    momentum_optimizer {
      learning_rate {
        manual_step_learning_rate {
          initial_learning_rate: 0.000199999994948
          schedule {
            step: 0
            learning_rate: 0.000199999994948
          }
          schedule {
            step: 900000
            learning_rate: 1.99999994948e-05
          }
          schedule {
            step: 1200000
            learning_rate: 1.99999999495e-06
          }
        }
      }
      momentum_optimizer_value: 0.899999976158
    }
    use_moving_average: false
  }
  gradient_clipping_by_norm: 10.0
  fine_tune_checkpoint: "PATH_TO_BE_CONFIGURED/model.ckpt"
  from_detection_checkpoint: true
  num_steps: 200000
}
train_input_reader {
  label_map_path: "PATH_TO_BE_CONFIGURED/mscoco_label_map.pbtxt"
  tf_record_input_reader {
    input_path: "PATH_TO_BE_CONFIGURED/mscoco_train.record"
  }
}
eval_config {
  num_examples: 8000
  max_evals: 10
  use_moving_averages: false
}
eval_input_reader {
  label_map_path: "PATH_TO_BE_CONFIGURED/mscoco_label_map.pbtxt"
  shuffle: false
  num_readers: 1
  tf_record_input_reader {
    input_path: "PATH_TO_BE_CONFIGURED/mscoco_val.record"
  }
}

 

the only advertisement that I see is "ILLEGAL INSTRUCTION" but I haven't found any log file to see details.

 

I Hope someone can help,

 

Thanks regards .

 

 

 

 

0 Kudos
Copy link
Reply

Community support is provided during standard business hours (Monday to Friday 7AM - 5PM PST). Other contact methods are available here.

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

For more complete information about compiler optimizations, see our Optimization Notice.