I've been working with Tensorflow for quite a while now, had some issues, but they never remained unresolved. Today i wanted to train a new model, when things got interesting. At first, the training stopped after one step without any reason. It happend before, opening a new cmd window solved it. Not this time tough. After i tried again, train.py started to throw this:
Windows fatal exception: access violation
Current thread 0x000018d4 (most recent call first):
File "C:\windows\system32\venv\lib\site-packages\tensorflow\python\lib\io\file_io.py", line 84 in _preread_check File "C:\windows\system32\venv\lib\site-packages\tensorflow\python\lib\io\file_io.py", line 122 in read File "C:\Users\xx\source\TensorFlow\models\research\object_detection\utils\label_map_util.py", line 133 in load_labelmap File "C:\Users\xx\source\TensorFlow\models\research\object_detection\utils\label_map_util.py", line 164 in get_label_map_dict File "C:\Users\xx\source\TensorFlow\models\research\object_detection\data_decoders\tf_example_decoder.py", line 59 in init File "C:\Users\xx\source\TensorFlow\models\research\object_detection\data_decoders\tf_example_decoder.py", line 314 in init File "C:\Users\xx\source\TensorFlow\models\research\object_detection\builders\dataset_builder.py", line 130 in build File "train.py", line 121 in get_next File "C:\Users\xx\source\TensorFlow\models\research\object_detection\legacy\trainer.py", line 59 in create_input_queue File "C:\Users\xx\source\TensorFlow\models\research\object_detection\legacy\trainer.py", line 280 in train File "train.py", line 180 in main File "C:\windows\system32\venv\lib\site-packages\tensorflow\python\util\deprecation.py", line 324 in new_func File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\lib\site-packages\absl\app.py", line 251 in _run_main File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\lib\site-packages\absl\app.py", line 300 in run File "C:\windows\system32\venv\lib\site-packages\tensorflow\python\platform\app.py", line 40 in run File "train.py", line 184 in
Last time i saw this issue, it was because i was using data downloaded from the internet, and there was one particular picture that TF did not like, but removing that one from the dataset solved the issue. I was wondering if this was the case, but no. I couldnt start it with previously tried datasets either... i decided to reinstall tensorflow, set up a new virtual environment, but still nothing. Been looking for hours what the problem could be, both on the internet, and on my own trying different things, but nothing worked, same exception each time. Did anybody encounter anything similar?
