pb生成pbtxt时出错

错误如下：

_graph_ssd.py --input frozen_inference_graph.pb --config ssd_mobilenet_v2_coco.config --output graph.pbtxt
Scale: [0.200000-0.950000]
Aspect ratios: [1.0, 2.0, 0.5, 3.0, 0.3333]
Reduce boxes in the lowest layer: True
Number of classes: 1
Number of layers: 6
box predictor: convolutional
Input image size: 300x300
Traceback (most recent call last):
  File "tf_text_graph_ssd.py", line 368, in <module>
    createSSDGraph(args.input, args.config, args.output)
  File "tf_text_graph_ssd.py", line 232, in createSSDGraph
    assert(graph_def.node[0].op == 'Placeholder')
AssertionError

解决方法

转换pb文件

import tensorflow as tf
from tensorflow.tools.graph_transforms import TransformGraph

with tf.gfile.FastGFile('frozen_inference_graph.pb', 'rb') as f:
    graph_def = tf.GraphDef()
    graph_def.ParseFromString(f.read())
    graph_def = TransformGraph(graph_def, ['image_tensor'], ['detection_boxes', 'detection_classes', 'detection_scores', 'num_detections'], ['sort_by_execution_order'])
    with tf.gfile.FastGFile('sorted_inference_graph.pb', 'wb') as f:
        f.write(graph_def.SerializeToString())

完美解决