search

aleksispi / drl-rpn-tf

Official Tensorflow implementation of drl-RPN: Deep Reinforcement Learning of Region Proposal Networks (CVPR 2018 paper)
MIT License
78 stars 21 forks source link

ImportError: No module named 'numpy.core.multiarray\r' #2

Closed wudongming97 closed 5 years ago

wudongming97 commented 5 years ago

Hello! I met some problems. Can you give me some suggestions? Thanks a lot!

`... Loaded datasetvoc_2007_trainvalfor training Set proposal method: gt Appending horizontally-flipped training examples... voc_2007_trainval gt roidb loaded from /home/cgv841/wudongming/drl-rpn-tf/data/cache/voc_2007_trainval_gt_roidb.pkl done Preparing training data... done Loaded datasetvoc_2012_trainval` for training Set proposal method: gt Appending horizontally-flipped training examples... Traceback (most recent call last): File "/home/cgv841/wudongming/drl-rpn-tf/tools/../lib/datasets/pascal_voc.py", line 111, in gt_roidb roidb = pickle.load(fid) ImportError: No module named 'numpy.core.multiarray\r'

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "./tools/trainval_net.py", line 117, in imdb, roidb = combined_roidb(args.imdb_name) File "./tools/trainval_net.py", line 89, in combined_roidb roidbs = [get_roidb(s) for s in imdb_names.split('+')] File "./tools/trainval_net.py", line 89, in roidbs = [get_roidb(s) for s in imdb_names.split('+')] File "./tools/trainval_net.py", line 86, in get_roidb roidb = get_training_roidb(imdb) File "/home/cgv841/wudongming/drl-rpn-tf/tools/../lib/model/train_val.py", line 460, in get_training_roidb imdb.append_flipped_images() File "/home/cgv841/wudongming/drl-rpn-tf/tools/../lib/datasets/imdb.py", line 114, in append_flipped_images boxes = self.roidb[i]['boxes'].copy() File "/home/cgv841/wudongming/drl-rpn-tf/tools/../lib/datasets/imdb.py", line 74, in roidb self._roidb = self.roidb_handler() File "/home/cgv841/wudongming/drl-rpn-tf/tools/../lib/datasets/pascal_voc.py", line 113, in gt_roidb roidb = pickle.load(fid, encoding='bytes') ImportError: No module named 'numpy.core.multiarray\r' Command exited with non-zero status 1 8.00user 0.71system 0:09.17elapsed 95%CPU (0avgtext+0avgdata 282404maxresident)k 0inputs+0outputs (0major+59566minor)pagefaults 0swaps

I run it on Ubuntu 16.04, python 3.5

In fact, in virtual env, I can import numpy.core.multiarray and the file independently. And, 'voc_2007_trainval gt roidb loaded from /home/cgv841/wudongming/drl-rpn-tf/data/cache/voc_2007_trainval_gt_roidb.pkl' is printed.

aleksispi commented 5 years ago

Hi wudongming97. Sorry for the late response. I have not encountered such an issue before. Do you get the same issue if you try to install and run the tf-faster-rcnn code on top of which drl-RPN is implemented? I.e. if you try to follow and run the steps in https://github.com/endernewton/tf-faster-rcnn? Because looking at your error message, it appears not to be related to any of the core drl-RPN code, but rather some dataset pickle stuff -- hence I suspect that the issue is at the level of tf-faster-rcnn (or even deeper down).

If you try that and it seems to work to use tf-faster-rcnn, then I'll investigate this more thoroughly as it has to do with the drl-RPN code somehow.

Btw, it should work on Ubuntu 16.04 with python 3.5.

Best, Aleksis

aleksispi commented 5 years ago

Closing due to inactivity -- likely an issue of a parent repo as mentioned in my above comment.