运行model_main_tf2.py报tensorflow compat模块无v2属性错误如何解决

报错场景 #

运行目标检测模型训练命令：

!python models/research/object_detection/model_main_tf2.py \
 --model_dir=Tensorflow/workspace/models/my_ssd_mobnet \
 --pipeline_config_path=Tensorflow/workspace/models/my_ssd_mobnet/pipeline.config \
 --num_training_steps=200

执行后抛出错误，核心报错信息为：
tensorflow._api.v1.compat.v2.compat' has no attribute 'v2

完整错误追踪栈：

Traceback (most recent call last):
  File "models/research/object_detection/model_main_tf2.py", line 113, in <module>
    tf.compat.v1.app.run()
  File "/tensorflow-1.15.2/python3.7/tensorflow_core/python/platform/app.py", line 40, in run
    _run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)
  File "/usr/local/lib/python3.7/dist-packages/absl/app.py", line 312, in run
    _run_main(main, args)
  File "/usr/local/lib/python3.7/dist-packages/absl/app.py", line 258, in _run_main
    sys.exit(main(argv))
  File "models/research/object_detection/model_main_tf2.py", line 101, in main
    strategy = tf.compat.v2.distribute.MirroredStrategy()
  File "/tensorflow-1.15.2/python3.7/tensorflow_core/python/util/module_wrapper.py", line 193, in __getattr__
    attr = getattr(self._tfmw_wrapped_module, name)
AttributeError: module 'tensorflow._api.v1.compat.v2.compat' has no attribute 'v2'

当前环境配置 #

当前配置TensorFlow Object Detection API的代码如下，环境强制指定使用TensorFlow 1.x版本：

%tensorflow_version 1.x
import os
import pathlib

# 若models仓库不存在则克隆
if "models" in pathlib.Path.cwd().parts:
  while "models" in pathlib.Path.cwd().parts:
    os.chdir('..')
elif not pathlib.Path('models').exists():
  !git clone --depth 1 https://github.com/cloud-annotations/models

!pip install cloud-annotations==0.0.4
!pip install tf_slim
!pip install lvis
!pip install --no-deps tensorflowjs==1.4.0

%cd /content/drive/MyDrive/Object_Detecation/models/research
!protoc object_detection/protos/*.proto --python_out=.

pwd = os.getcwd()
os.environ['PYTHONPATH'] += f':{pwd}:{pwd}/slim'

!python object_detection/builders/model_builder_tf2_test.py

报错根因：版本不匹配。当前环境为TensorFlow 1.x，却运行了仅支持TensorFlow 2.x的训练脚本model_main_tf2.py，TF1.x的compat.v2模块不包含脚本调用的TF2专属分布式策略接口，最终触发属性不存在错误。

解决方案 #

二选一即可：

• 方案1（推荐，匹配现有TF2训练脚本）：切换到TensorFlow 2.x环境
  1. 将配置首行的%tensorflow_version 1.x替换为%tensorflow_version 2.x，本地环境则直接卸载TF1.x，安装2.5~2.10区间的稳定TF2版本（该区间对Object Detection API兼容性最佳）
  2. 重新执行依赖安装流程，移除tensorflowjs==1.4.0的旧版本限制，安装适配TF2的tensorflowjs版本
  3. 重新执行proto编译、PYTHONPATH配置步骤，运行model_builder_tf2_test.py确认所有测试用例通过后，再执行训练命令即可
• 方案2（保留TF1.x环境）：替换为TF1兼容的训练脚本
  1. 不要运行model_main_tf2.py，改用同目录下TF1版本的训练脚本model_main.py
  2. 替换pipeline配置文件为TF1兼容格式，不可直接复用TF2版本的pipeline.config
  3. 重新走TF1版本的API验证流程后再启动训练

内容的提问来源于stack exchange，提问作者None -