pengxj / action-faster-rcnn Goto Github PK

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

I scan the listfile about JHMDB.Why filename in JHMDB_FLOW_5_split_0.trainlist is started frome 000003.jpg?
If I want train with my dataset,which format in trainlist should I use?

Hi,@pengxj, thanks for the work, I just reviewed the paper on the multi-region part, which has 4 parts including the whole body, upper body, bottom body, but I don't know how is the 4th region defined, excluding the body part? Additionally for the multi-region implementation, I thought it would be implemented directly on the roi_data_layer/minibatch.py or roi_data_layer/layer.py, but I didn't find there is connection, could you tell me where this part is being implemented ?

Hi,
I am trying to compare your work with my "action detection" work. I need few details of your work.
Can you provide us the below details of your model:

1. Number of trainable parameters
2. G- FLOP's
3. Inference fps

When I try this cmmond :
python action_tools/eval_linked_results.py --imdb UCF101_RGB_1_FLOW_5_split_0 --res ucf101_vdets_3scales_rgb1flow5.pkl

Error shows:
IOError: [Errno 2] No such file or directory: '/home/guan/github/action-faster-rcnn/data/UCF101/UCF101_24_Annotations/Basketball/v_Basketball_g01_c01.xgtf'

Could you teach me how to solve this problem?

Hi @pengxj,

When I run the following command
"./action_experiments/scripts/train_action_det.sh 0 VGG_16 UCF101 5 FLOW 0"
I got an error:
"F0609 23:08:41.226728 14455 io.cpp:54] Check failed: fd != -1 (-1 vs. -1) File not found: models/imagenet_models/5_VGG_16.caffemodel"
It seems that there should be a pre-trained 5 stacked frames model file. Can you share this file? Thanks!

Hi @pengxj ,
It seems that the cudnn v5 is not supported in this project, When I make with cudnn v5, it reports errors, while it's fine using cudnn v4. Could you please update the upstream of caffe to make it support cudnn v5? thanks.

Hi, I have downloaded the ucf-sport dataset and I can not find the humans.txt and videos.txt, so would you please explain what those txts stands for ?

I've read your paper and it inpaired me a lot. I want to follow your work. I downloaded UCFsport dataset but I cannot find video.txt and the name of the dateset's images and videos are different from your filelist. So can you provide your dataset's file stucture. There are similar problems when I was trying the UCF101 dataset. Thank you.