Reputation: 25
I am using a pretrained model by TensorFlow , Attention ocr https://github.com/tensorflow/models/tree/master/research/attention_ocr
While running eval.py the script always runs into an error
File "C:\Users\DELL\Anaconda3\envs\python35DL\lib\site-packages\tensorflow\python\framework\errors_impl.py", line 519, in __exit__
c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.NotFoundError: NewRandomAccessFile failed to Create/Open: D:\Vision_Project\attention_ocr\python\data/fsns\charset_size=134.txt : The system cannot find the path specified.
; No such process
I am unable to understand which path it cannot find , where and how do I fix this in the script .
System Configurations (if it matters) : Windows 10 , 8gb RAM , Core i3 , 1 TB HDD Tensorflow Version : 1.8.0 Python Version : 3.5 running through VirtualEnv
Upvotes: 2
Views: 7072
Reputation: 947
I had a similar problem.. the system can't find the specific files.
After I print out the exact file name string read by the system, I found out that there are some '\n' in the file name...
Problem solved after I do line.rstrip() .
Upvotes: 1
Reputation: 1238
It is a bug - fsns.py
uses Linux separator for a part of the path. Here is the fix.
Please try and let me know if it helped. Hopefully it will be merged soon.
Upvotes: 1