About the problem of pickle

Question

About the problem of pickle

Closed this issue 7 years ago · 11 comments

Traceback (most recent call last):
File "A2RL.py", line 15, in
var_dict = pickle.load(f)
File "/home/stu_3/anaconda2/lib/python2.7/pickle.py", line 1384, in load
return Unpickler(file).load()
File "/home/stu_3/anaconda2/lib/python2.7/pickle.py", line 864, in load
dispatchkey
KeyError: '<'

I have met this problem when running the sh example.sh file. How to solve this problem? Thank you very much!

Answer 1 · 2017-10-14T11:15:36.000Z

The code is only tested with python3, try to run it with python 3.5 or python 3.6

Answer 2 · 2017-10-14T11:55:46.000Z

@wuhuikai
I installed python3 with Anaconda (also with python2.7), and utilize the conda create --name python3 python=3.6 activate python3. But the same problem still exists.

(python3) stu_3@rs:~/Documents/task14/DRL/DRL_img/TF-A2RL$ sh example.sh
Traceback (most recent call last):
File "A2RL.py", line 15, in
var_dict = pickle.load(f)
_pickle.UnpicklingError: invalid load key, '<'.

Answer 3 · 2017-10-14T12:09:25.000Z

@whuhxb I run the code as your instructions and met no errors. A possible cause is that the vfn_rl.pkl is broken. The md5sum of vfn_rl.pkl is f40cc5de88aaf72ae930ac23ba030a92, what about yours ?

Answer 4 · 2017-10-14T12:38:24.000Z

@wuhuikai
I download the vfn_rl.pkl again, the previous problem disappeared. I want to know whether your code only supports the jpg format. Does it support the tiff format images? In addition, how to utilize your code for batch processing because I have a large number of images to be processed.

Answer 5 · 2017-10-14T12:42:26.000Z

@whuhxb It support all formats supported by scikit-image.
If you want to batch process images, you can look the example code Here.

Answer 6 · 2017-10-14T12:43:18.000Z

@wuhuikai
Another question: Can your code directly be utilized on my own data with your vfn_rl.pkl? Or should I train my own data with your code by obtaining the vfn_rl.pkl?

Answer 7 · 2017-10-14T12:47:46.000Z

@whuhxb It should generalize to your data well, but I'm not sure. Currently the training code is not available.

Answer 8 · 2017-10-14T12:49:15.000Z

@wuhuikai
Thank you very much. If I have several images in an folder to be cropped, in order to utilize the batch processing, should I just give the --image_path and --save_path without the specific .jpg images?

Answer 9 · 2017-10-14T12:49:27.000Z

@whuhxb You shouldn't. You'd better write your own main function to read a batch of images from files, feed them to a2rl as a list and then crop the images and save them.

Answer 10 · 2017-10-14T12:51:20.000Z

Since your main issue has been resolved, I decide to close this issue.

Answer 11 · 2017-10-14T12:52:35.000Z

@wuhuikai
OK. Thank you very much!