pengxj / action-faster-rcnn

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ImportError: cannot import name im_detect_2stream

NIEYALI opened this issue · comments

commented

Thanks for your great work, but when I tried test reuslt, I met the following problem, Do you have any idea?
Thanks a lot @vra
python action_tools/eval_linked_results.py --imdb UCF101_RGB_1_FLOW_5_split_0 --res /home/user1/YALI/action-faster-rcnn/ucf101_vdets_3scales_rgb1flow5.pkl
Traceback (most recent call last):
File "action_tools/eval_linked_results.py", line 5, in
import action_util as action
File "/home/user1/YALI/action-faster-rcnn/action_tools/action_util.py", line 8, in
from fast_rcnn.test import im_detect, im_detect_2stream
ImportError: cannot import name im_detect_2stream

Hi @NIEYALI,
This error appears because the define of function im_detect_2stream is not found in file ROOT/lib/fast_rcnn/test.py, I met this error too when I ran the code. So this is an error of source code, waiting for the answer from author @pengxj.
As a trial, I change the Line 8 to from fast_rcnn.test import im_detect, and Line 112 to scores, boxes = im_detect(net, im, flows), this error disappears. But I don't the difference between im_detect and im_detect_2stream, hoping to hear from author.

commented

Thanks for your reply, But I have another question following:
if 'RGB' in image_set and 'FLOW' in image_set:
self._data_path = '/home/lear/xpeng/data/UCF101/flows_color'
else:
self.MOD = image_set.split('')[1]
self.LEN = image_set.split('')[2]
self._data_path = None
if self._MOD=='RGB': self._data_path = '/home/lear/xpeng/data/UCF101/frames_240'
if self._MOD=='FLOW': self._data_path = '/home/lear/xpeng/data/UCF101/flows_color'

These three path, where could I find' flows_color, frames_240, flows_color'
I have no idea about them, could you give some suggestions?
Thanks a lot.
@vra
@pengxj

Hi @NIEYALI ,
These paths are where you host your RGB images and flow images, which is not included in the source code, so you may change them to the paths to image data in your computer.

Hi @NIEYALI @vra ,
I just updated the test.py file in lib/fast_rcnn. Now should be ok.

Hi @vra @NIEYALI ,
The code for optical flow is updated, thanks for checking.

Hi @pengxj,
Thanks for your quick updating! 👍

Hi @pengxj @vra

I dowdload the UCF101 datasets,but i cannot find the anno files, could you gave me the link that i can download the files,for example baidupan or driverbox?

Hi @vra

self._data_path = '/home/lear/xpeng/data/UCF101/flows_color'
......
if self._MOD=='RGB': self._data_path = '/home/lear/xpeng/data/UCF101/frames_240'
if self._MOD=='FLOW': self._data_path = '/home/lear/xpeng/data/UCF101/flows_color'

what is these three path stand for?

@chengshuai Please refer to the THUMOS Challenge 2015 website for annotations or you can find some information here https://github.com/gurkirt/corrected-UCF101-Annots.

'frames_240' is the root dir. where you store your videos frames of the dataset. And flows_color is the root of flow frames.