TensorFlow模型加载遇OpError错误：图形识别网站部署求助

问题描述 #

正在搭建图形识别网站，支持用户上传矩形、平行四边形、正方形、菱形或梯形图片，经AI识别后在HTML页面返回结果。但执行以下代码加载模型时触发OpError：

model_path = './Model/Model'
model = tf.saved_model.load(model_path)

完整报错栈如下：

Traceback (most recent call last): File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\flask\app.py", line 2552, in __call__ return self.wsgi_app(environ, start_response) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\flask\app.py", line 2532, in wsgi_app response = self.handle_exception(e) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\flask\app.py", line 2529, in wsgi_app response = self.full_dispatch_request() File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\flask\app.py", line 1825, in full_dispatch_request rv = self.handle_user_exception(e) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\flask\app.py", line 1823, in full_dispatch_request rv = self.dispatch_request() File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\flask\app.py", line 1799, in dispatch_request return self.ensure_sync(self.view_functions[rule.endpoint])(**view_args) File "c:\Users\내문서\Desktop\박민찬\@상일고 박민찬 보고서 및 자료\1학년\@ 자율교육과정, 담다프로젝트 모음\2학기\@ 2학기 자율교육과정_Machine Learning 개발과정\모델 활용 웹사이트\main.py", line 49, in upload_image      predictions = predict_image(filename) File "c:\Users\내문서\Desktop\박민찬\@상일고 박민찬 보고서 및 자료\1학년\@ 자율교육과정, 담다프로젝트 모음\2학기\@ 2학기 자율교육과정_Machine Learning 개발과정\모델 활용 웹사이트\main.py", line 31, in predict_image     model = tf.saved_model.load(model_path) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\tensorflow\python\saved_model\load.py", line 828, in load result = load_partial(export_dir, None, tags, options)["root"] File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\tensorflow\python\saved_model\load.py", line 959, in load_partial ckpt_options, options, filters) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\tensorflow\python\saved_model\load.py", line 192, in __init__ self._restore_checkpoint() File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\tensorflow\python\saved_model\load.py", line 527, in _restore_checkpoint load_status = saver.restore(variables_path, self._checkpoint_options) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\tensorflow\python\checkpoint\checkpoint.py", line 1459, in restore object_graph_string = reader.get_tensor(base.OBJECT_GRAPH_PROTO_KEY) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\tensorflow\python\training\py_checkpoint_reader.py", line 71, in get_tensor error_translator(e) File "C:\Users\내문서\AppData\Local\Programs\Python\Python37\lib\site-packages\tensorflow\python\training\py_checkpoint_reader.py", line 45, in error_translator raise errors_impl.OpError(None, None, error_message, errors_impl.UNKNOWN) tensorflow.python.framework.errors_impl.OpError

需求是让模型正常运行，在HTML页面返回识别结果。

排查与解决步骤 #

1. 检查模型文件完整性 #

• 确认./Model/Model目录下存在saved_model.pb文件，以及包含variables.data-00000-of-00001和variables.index的variables文件夹，缺失任一文件都会导致加载失败。
• 若文件损坏，回到训练环境，用tf.saved_model.save(model, export_dir)重新导出完整模型，再复制到项目目录。

2. 验证TensorFlow版本兼容性 #

• 查看训练模型时使用的TensorFlow版本，确保运行环境的版本与训练版本主版本号一致，次版本号尽量接近（跨大版本如TF1.x和TF2.x会直接报错）。
• 执行pip install tensorflow==[对应版本号]调整运行环境版本。

3. 修正文件路径问题 #

• Flask默认以启动脚本所在目录为工作目录，相对路径可能因运行位置出错，改用绝对路径测试：

  import os
  model_path = os.path.join(os.path.dirname(__file__), 'Model', 'Model')
  model = tf.saved_model.load(model_path)
  

4. 检查文件读取权限 #

• 确保运行Python的用户对./Model/Model及内部文件有读取权限：Windows下右键文件夹→属性→安全，确认当前用户权限；Linux/macOS下执行chmod -R 755 ./Model。

5. 尝试Keras加载方式 #

• 若模型是用Keras的model.save()导出的，改用tf.keras.models.load_model()加载：

  model = tf.keras.models.load_model(model_path)
  

内容的提问来源于stack exchange，提问作者bkmnchn