TF object detection API detection model retraining: "object_detection.protos.SsdFeatureExtractor" has no field named "batch_norm_trainable"

Question

I'm retraining mobilenet v2 SSD using Tensorflow Object Detection API and I get this particular error on both Windows and Ubuntu environments.
My environment below is Windows 10 with python 3.5, cuda 9.0, and tensorflow-gpu 1.6. I compiled protobuf with protobuf 3.4.0.
I did the installation test and it was okay, so now I'm trying with my own dataset and getting google.protobuf.text_format.ParseError: 108:7 : Message type "object_detection.protos.SsdFeatureExtractor" has no field named "batch_norm_trainable". error.

Things I tried:

• Using tf 1.5 - 1.7
• Training mobilenet v1 SSD

The whole error code:

C:\Users\kazuy\models\research>python object_detection/train.py --logtostderr --train_dir="C:\\Users\\kazuy\\models\\research\\object_detection\\projects\\coco_selfcar\\tests\\test30" --pipeline_config_path="C:\\Users\\kazuy\\models\\research\\object_detection\\projects\\coco_selfcar\\ssd_mobilenet_v2_coco.config"
Traceback (most recent call last):
  File "object_detection/train.py", line 167, in <module>
    tf.app.run()
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\platform\app.py", line 126, in run
    _sys.exit(main(argv))
  File "object_detection/train.py", line 92, in main
    FLAGS.pipeline_config_path)
  File "C:\Users\kazuy\models\research\object_detection\utils\config_util.py", line 93, in get_configs_from_pipeline_file
    text_format.Merge(proto_str, pipeline_config)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 533, in Merge
    descriptor_pool=descriptor_pool)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 587, in MergeLines
    return parser.MergeLines(lines, message)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 620, in MergeLines
    self._ParseOrMerge(lines, message)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 635, in _ParseOrMerge
    self._MergeField(tokenizer, message)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 735, in _MergeField
    merger(tokenizer, message, field)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 823, in _MergeMessageField
    self._MergeField(tokenizer, sub_message)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 735, in _MergeField
    merger(tokenizer, message, field)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 823, in _MergeMessageField
    self._MergeField(tokenizer, sub_message)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 735, in _MergeField
    merger(tokenizer, message, field)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 823, in _MergeMessageField
    self._MergeField(tokenizer, sub_message)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\google\protobuf\text_format.py", line 703, in _MergeField
    (message_descriptor.full_name, name))
google.protobuf.text_format.ParseError: 108:7 : Message type "object_detection.protos.SsdFeatureExtractor" has no field named "batch_norm_trainable".

The config file that I use is mostly unchanged from the sample files from here:

# SSD with Mobilenet v2 configuration for MSCOCO Dataset.
# Users should configure the fine_tune_checkpoint field in the train config as
# well as the label_map_path and input_path fields in the train_input_reader and
# eval_input_reader. Search for "PATH_TO_BE_CONFIGURED" to find the fields that
# should be configured.

model {
  ssd {
    num_classes:3
    box_coder {
      faster_rcnn_box_coder {
        y_scale: 10.0
        x_scale: 10.0
        height_scale: 5.0
        width_scale: 5.0
      }
    }
    matcher {
      argmax_matcher {
        matched_threshold: 0.5
        unmatched_threshold: 0.5
        ignore_thresholds: false
        negatives_lower_than_unmatched: true
        force_match_for_each_row: true
      }
    }
    similarity_calculator {
      iou_similarity {
      }
    }
    anchor_generator {
      ssd_anchor_generator {
        num_layers: 6
        min_scale: 0.2
        max_scale: 0.95
        aspect_ratios: 1.0
        aspect_ratios: 2.0
        aspect_ratios: 0.5
        aspect_ratios: 3.0
        aspect_ratios: 0.3333
      }
    }
    image_resizer {
      fixed_shape_resizer {
        height: 300
        width: 300
      }
    }
    box_predictor {
      convolutional_box_predictor {
        min_depth: 0
        max_depth: 0
        num_layers_before_predictor: 0
        use_dropout: false
        dropout_keep_probability: 0.8
        kernel_size: 3
        box_code_size: 4
        apply_sigmoid_to_scores: false
        conv_hyperparams {
          activation: RELU_6,
          regularizer {
            l2_regularizer {
              weight: 0.00004
            }
          }
          initializer {
            truncated_normal_initializer {
              stddev: 0.03
              mean: 0.0
            }
          }
          batch_norm {
            train: true,
            scale: true,
            center: true,
            decay: 0.9997,
            epsilon: 0.001,
          }
        }
      }
    }
    feature_extractor {
      type: 'ssd_mobilenet_v2'
      min_depth: 16
      depth_multiplier: 1.0
      use_depthwise: true
      conv_hyperparams {
        activation: RELU_6,
        regularizer {
          l2_regularizer {
            weight: 0.00004
          }
        }
        initializer {
          truncated_normal_initializer {
            stddev: 0.03
            mean: 0.0
          }
        }
        batch_norm {
          train: true,
          scale: true,
          center: true,
          decay: 0.9997,
          epsilon: 0.001,
        }
      }
      batch_norm_trainable: true
    }
    loss {
      classification_loss {
        weighted_sigmoid {
        }
      }
      localization_loss {
        weighted_smooth_l1 {
        }
      }
      hard_example_miner {
        num_hard_examples: 3000
        iou_threshold: 0.99
        loss_type: CLASSIFICATION
        max_negatives_per_positive: 3
        min_negatives_per_image: 3
      }
      classification_weight: 1.0
      localization_weight: 1.0
    }
    normalize_loss_by_num_matches: true
    post_processing {
      batch_non_max_suppression {
        score_threshold: 1e-8
        iou_threshold: 0.6
        max_detections_per_class: 100
        max_total_detections: 100
      }
      score_converter: SIGMOID
    }
  }
}

train_config: {
  batch_size: 24
  optimizer {
    rms_prop_optimizer: {
      learning_rate: {
        exponential_decay_learning_rate {
          initial_learning_rate: 0.004
          decay_steps: 800720
          decay_factor: 0.95
        }
      }
      momentum_optimizer_value: 0.9
      decay: 0.9
      epsilon: 1.0
    }
  }
  fine_tune_checkpoint: "C:\\Users\\kazuy\\models\\research\\object_detection\\model_zoo\\ssd_mobilenet_v2_coco_2018_03_29\\model.ckpt"
  fine_tune_checkpoint_type:  "detection"
  # Note: The below line limits the training process to 200K steps, which we
  # empirically found to be sufficient enough to train the pets dataset. This
  # effectively bypasses the learning rate schedule (the learning rate will
  # never decay). Remove the below line to train indefinitely.
  num_steps: 200000
  data_augmentation_options {
    random_horizontal_flip {
    }
  }
  data_augmentation_options {
    ssd_random_crop {
    }
  }
}

train_input_reader: {
  tf_record_input_reader {
    input_path: "C:\\Users\\kazuy\\models\\research\\object_detection\\projects\\coco_selfcar\\tests\\test30\\train.tfrecords"
  }
  label_map_path: "C:\\Users\\kazuy\\models\\research\\object_detection\\projects\\coco_selfcar\\label_map.pbtxt"
}

eval_config: {
  num_examples: 8000
  # Note: The below line limits the evaluation process to 10 evaluations.
  # Remove the below line to evaluate indefinitely.
  max_evals: 10
}

eval_input_reader: {
  tf_record_input_reader {
    input_path: "PATH_TO_BE_CONFIGURED/mscoco_val.record"
  }
  label_map_path: "PATH_TO_BE_CONFIGURED/mscoco_label_map.pbtxt"
  shuffle: false
  num_readers: 1
}

And this is the labelmap:

item {
  id: 1
  name: 'Pedestrian'
}

item {
  id: 2
  name: 'Car'
}

item {
  id: 3
  name: 'Truck'
}

When I try the training with mobilenet v1 SSD by using a modified version of this config file I get a different error:

C:\Users\kazuy\models\research>python object_detection/train.py --logtostderr --train_dir="C:\\Users\\kazuy\\models\\research\\object_detection\\projects\\coco_selfcar\\tests\\test30" --pipeline_config_path="C:\\Users\\kazuy\\models\\research\\object_detection\\projects\\coco_selfcar\\ssd_mobilenet_v1_pets.config"
WARNING:tensorflow:From C:\Users\kazuy\models\research\object_detection\trainer.py:257: create_global_step (from tensorflow.contrib.framework.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Please switch to tf.train.create_global_step
Traceback (most recent call last):
  File "object_detection/train.py", line 167, in <module>
    tf.app.run()
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\platform\app.py", line 126, in run
    _sys.exit(main(argv))
  File "object_detection/train.py", line 163, in main
    worker_job_name, is_chief, FLAGS.train_dir)
  File "C:\Users\kazuy\models\research\object_detection\trainer.py", line 275, in train
    clones = model_deploy.create_clones(deploy_config, model_fn, [input_queue])
  File "C:\Users\kazuy\models\research\slim\deployment\model_deploy.py", line 193, in create_clones
    outputs = model_fn(*args, **kwargs)
  File "C:\Users\kazuy\models\research\object_detection\trainer.py", line 198, in _create_losses
    prediction_dict = detection_model.predict(images, true_image_shapes)
  File "C:\Users\kazuy\models\research\object_detection\meta_architectures\ssd_meta_arch.py", line 384, in predict
    preprocessed_inputs)
  File "C:\Users\kazuy\models\research\object_detection\models\ssd_mobilenet_v1_feature_extractor.py", line 121, in extract_features
    scope=scope)
  File "C:\Users\kazuy\models\research\slim\nets\mobilenet_v1.py", line 267, in mobilenet_v1_base
    scope=end_point)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\contrib\framework\python\ops\arg_scope.py", line 183, in func_with_args
    return func(*args, **current_args)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\contrib\layers\python\layers\layers.py", line 1059, in convolution
    outputs = normalizer_fn(outputs, **normalizer_params)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\contrib\framework\python\ops\arg_scope.py", line 183, in func_with_args
    return func(*args, **current_args)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\contrib\layers\python\layers\layers.py", line 651, in batch_norm
    outputs = layer.apply(inputs, training=is_training)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\layers\base.py", line 809, in apply
    return self.__call__(inputs, *args, **kwargs)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\layers\base.py", line 696, in __call__
    outputs = self.call(inputs, *args, **kwargs)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\layers\normalization.py", line 544, in call
    training_value = utils.constant_value(training)
  File "C:\Users\kazuy\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\layers\utils.py", line 240, in constant_value
    raise TypeError('`pred` must be a Tensor, a Variable, or a Python bool.')

I have been successfully training mobilenetv1 ssd models on ubuntu last month, and I started encountering this error after installing the repository from scratch today.
I'd appreciate any help! Thank you!

Answer 1

For ssd mobilenet v2, you could remove "batch_norm_trainable: true" as this field is deprecated now.