Tensorflow对象检测API检查点读取器值错误

Question

我已经跟踪了这个{a1}和这个{a2}。 因此，我：

• tensorflow gpu==2.3.1
• 正确安装对象检测
• 遵守API文档的工作目录

但当我尝试运行时：model_main_tf2.py

我得到一个错误：

  File "defects_detection\model_main_tf2.py", line 113, in <module>
    tf.compat.v1.app.run()
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\tensorflow\python\platform\app.py", line 40, in run
    _run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\absl\app.py", line 300, in run
    _run_main(main, args)
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\absl\app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "defects_detection\model_main_tf2.py", line 110, in main
    record_summaries=FLAGS.record_summaries)
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\object_detection\model_lib_v2.py", line 566, in train_loop
    unpad_groundtruth_tensors)
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\object_detection\model_lib_v2.py", line 339, in load_fine_tune_checkpoint
    if not is_object_based_checkpoint(checkpoint_path):
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\object_detection\model_lib_v2.py", line 302, in is_object_based_checkpoint
    var_names = [var[0] for var in tf.train.list_variables(checkpoint_path)]
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\tensorflow\python\training\checkpoint_utils.py", line 98, in list_variables
    reader = load_checkpoint(ckpt_dir_or_file)
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\tensorflow\python\training\checkpoint_utils.py", line 67, in load_checkpoint
    return py_checkpoint_reader.NewCheckpointReader(filename)
  File "G:\Documents\Florian\Programmation\DIADES_Defects_Detection\venv\lib\site-packages\tensorflow\python\training\py_checkpoint_reader.py", line 96, in NewCheckpointReader
    return CheckpointReader(a)
ValueError

它们不是ValueError之后的文本

我读了又读了我的pipeline.config。看起来不错。 我在网上查。我没有找到解决办法。 你对这个问题有什么建议或解决办法吗

多谢各位

Answer 1

所以我终于找到了问题所在。 我忘记了pipeline.congif路径中的“s”代表“fine\u tune\u checkpoint”

因此，我们将知道，此错误意味着您在检查点路径中出错