Open PCIHD opened 5 years ago
Hi, check this link https://youtu.be/nZUxoHPFf4w
change the attribute step: 0 to step: 1 in the optimizer section of the pipeline.config file. my guess is that it was changed in newer tensor flow object detector versions (training starts at step 1).
I am facing this issue saying first step shouldnt be zero , now if change is made to the config file in the config file at line 92 a new problem arises the model doesnt train at all
Traceback (most recent call last): File "train.py", line 184, in
tf.app.run()
File "D:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\platform\app.py", line 125, in run
_sys.exit(main(argv))
File "D:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\util\deprecation.py", line 324, in new_func
return func(*args, **kwargs)
File "train.py", line 180, in main
graph_hook_fn=graph_rewriter_fn)
File "D:\ProgramData\Anaconda3\lib\site-packages\object_detection-0.1-py3.7.egg\object_detection\legacy\trainer.py", line 304, in train
train_config.optimizer)
File "D:\ProgramData\Anaconda3\lib\site-packages\object_detection-0.1-py3.7.egg\object_detection\builders\optimizer_builder.py", line 52, in build
learning_rate = _create_learning_rate(config.learning_rate)
File "D:\ProgramData\Anaconda3\lib\site-packages\object_detection-0.1-py3.7.egg\object_detection\builders\optimizer_builder.py", line 115, in _create_learning_rate
learning_rate_sequence, config.warmup)
File "D:\ProgramData\Anaconda3\lib\site-packages\object_detection-0.1-py3.7.egg\object_detection\utils\learning_schedules.py", line 160, in manual_stepping
raise ValueError('First step cannot be zero.')
ValueError: First step cannot be zero.