Question

在Colab笔记本中运行model_main.py脚本时出现以下错误。

I0615 10:47:44.585545 139791270102912 basic_session_run_hooks.py:262] loss = 36.265278, step = 0
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/client/session.py", line 1365, in _do_call
    return fn(*args)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/client/session.py", line 1350, in _run_fn
    target_list, run_metadata)
  File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/client/session.py", line 1443, in _call_tf_sessionrun
    run_metadata)
tensorflow.python.framework.errors_impl.InvalidArgumentError: {{function_node __inference_Dataset_map_transform_and_pad_input_data_fn_1404}} assertion failed: [[0.3875]] [[0.375]]
     [[{{node Assert/AssertGuard/else/_25/Assert}}]]
     [[IteratorGetNext]]

有时模型训练最多进行1000步，而其他时候则立即失败。

我一直在尝试研究错误function_node __inference_Dataset_map_transform_and_pad_input_data_fn_1404，但无济于事。

我曾尝试对 pipeline.config 进行各种更改，尽管之前的尝试是在另一种经过预先训练的模型上进行的，但之前的尝试已经成功地完成了几千步。

有什么想法吗？

这是管道

<code>
model {
  ssd {
    num_classes: 2
    image_resizer {
      fixed_shape_resizer {
        height: 320
        width: 320
      }
    }

    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
        use_matmul_gather: false
      }
    }
    similarity_calculator {
      iou_similarity {
      }
    }
    anchor_generator {
      ssd_anchor_generator {
        num_layers: 6
        min_scale: 0.1
        max_scale: 0.5
        aspect_ratios: 0.3333
      }
    }

   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 {
            decay: 0.999700009823
            center: true
            scale: true
            epsilon: 0.0010000000475
            train: true
          }
        }
        min_depth: 0
        max_depth: 0
        num_layers_before_predictor: 0
        use_dropout: false
        kernel_size: 3
        box_code_size: 4
        apply_sigmoid_to_scores: false
      }
    }
     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,
        }
      }
   override_base_feature_extractor_hyperparams: true
    }
    loss {
      classification_loss {
        weighted_sigmoid {
        }
      }

    }
    normalize_loss_by_num_matches: true
    loss {
      localization_loss {
        weighted_smooth_l1 {
        }
      }
      classification_loss {
        weighted_sigmoid {
        }
      }
      hard_example_miner {
        num_hard_examples: 3000
        iou_threshold: 0.990000009537
        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: 5
        max_total_detections: 10
      }

  }
  }
}
train_config {
  batch_size: 8
  data_augmentation_options {
    random_horizontal_flip {
      keypoint_flip_permutation: 1
      keypoint_flip_permutation: 0
      keypoint_flip_permutation: 2
      keypoint_flip_permutation: 3
      keypoint_flip_permutation: 5
      keypoint_flip_permutation: 4
    }
  }

  optimizer {
    rms_prop_optimizer {
      learning_rate {
        exponential_decay_learning_rate {
          initial_learning_rate: 0.00400000018999
          decay_steps: 800720
          decay_factor: 0.949999988079
        }
      }
      momentum_optimizer_value: 0.899999976158
      decay: 0.899999976158
      epsilon: 1.0
    }
  }
  fine_tune_checkpoint: "/content/models/facessd_mobilenet_v2_quantized_320x320_open_image_v4/model.ckpt"
  fine_tune_checkpoint_type:  "detection"
  from_detection_checkpoint: true
  num_steps: 50000

}
train_input_reader {
  label_map_path: "/content/tensorflow/workspace/dataset/prepared/record/label_map.pbtxt"
  tf_record_input_reader {
    input_path: "/content/tensorflow/workspace/dataset/prepared/record/train.record"
  }
}
eval_config {
  metrics_set: "coco_detection_metrics"
  use_moving_averages: false
  num_examples: 185
  max_evals: 10
}
eval_input_reader {
  label_map_path: "/content/tensorflow/workspace/dataset/prepared/record/label_map.pbtxt"
  shuffle: false
  num_readers: 1
  tf_record_input_reader {
    input_path: "/content/tensorflow/workspace/dataset/prepared/record/test.record"
  }
}
graph_rewriter {
  quantization {
    delay: 500000
    weight_bits: 8
    activation_bits: 8
  }
}

Answer 1

事实证明，训练数据的边界框中存在错误。我更正了这些错误，现在可以进行培训。始终检查您的数据！

Tensorflow对象检测API训练错误

1 个答案: