Comments (17)
i've checked log file by executing script_faster_rcnn_demo
and i figure that the last two line of log file says
F0106 19:55:17.793050 7656 syncedmem.cpp:51] Check failed: error == cudaSuccess (2 vs. 0) out of memory
F0106 19:55:17.793050 7656 syncedmem.cpp:51] Check failed: error == cudaSuccess (2 vs. 0) out of memory
it says 'out of memory'.
is it because of my gpu? My gpu is gtx 650 ti.
if gpu is under Titan or tesla (ex. T20 ,T40 etc..), does it end before start?
i mean my running ends without this part below.
001763.jpg (500x375): time 0.201s (resize+conv+proposal: 0.150s, nms+regionwise: 0.052s)
004545.jpg (500x375): time 0.201s (resize+conv+proposal: 0.151s, nms+regionwise: 0.050s)
000542.jpg (500x375): time 0.192s (resize+conv+proposal: 0.151s, nms+regionwise: 0.041s)
from faster_rcnn.
I have similar problem as yours, anyone had solved this problem?
///////////////////////////////////////////////////////////////////////////
Error using caffe_
glog check error, please check log and clear mex
Error in caffe.Net/reshape (line 171)
caffe_('net_reshape', self.hNet_self);
Error in caffe.Net/reshape_as_input (line 186)
self.reshape();
Error in fast_rcnn_train (line 112)
caffe_solver.net.reshape_as_input(net_inputs);
Error in Faster_RCNN_Train.do_fast_rcnn_train (line 7)
model_stage.output_model_file = fast_rcnn_train(conf, dataset.imdb_train, dataset.roidb_train, ...
Error in script_faster_rcnn_pku (line 54)
model.stage1_fast_rcnn = Faster_RCNN_Train.do_fast_rcnn_train(conf_fast_rcnn, dataset, model.stage1_fast_rcnn, opts.do_val);
from faster_rcnn.
@Cogito2012
may i ask what your gpu is?
mine is gtx 650 ti
from faster_rcnn.
@Anylee2142 My GPU is Quadro K620, its only 2G memory
from faster_rcnn.
@Cogito2012
did you run
experiments/script_faster_rcnn_demo?
i thought my problem that i cannot run script_faster_rcnn_demo was because of my gpu.
but i googled and figured out that the gap between K620 and 650 ti is not that huge.
what's wrong with me..
can you tell me your computer specification?
from faster_rcnn.
My computer is a workstation T1700
My errors happened when I run training file experiments/script_faster_rcnn_ZF0712.m,
And I can successfully run the test file experiments/script_faster_rcnn_demo.m, you can try to modify the configs in /script_faster_rcnn_demo.m:
opts.use_gpu = false;
or use the ZF net:
% model_dir = fullfile(pwd, 'output', 'faster_rcnn_final', 'faster_rcnn_VOC0712_vgg_16layers'); %% VGG-16
model_dir = fullfile(pwd, 'output', 'faster_rcnn_final', 'faster_rcnn_VOC0712_ZF'); %% ZF
from faster_rcnn.
thank you for your advice. i can run demo now !
and if i may, why ' opts.use_gpu = false; ' make it run?
i mean obviously faster r-cnn code needs gpu to be runned.
Doesn't ' opts.use_gpu = false; ' this mean not using gpu?
Do you know difference between opts.use_gpu = true and opts.use_gpu = false ?
from faster_rcnn.
I can also run demo with opts.use_gpu = true only for ZF net.
you can read the README.MD:
Requirements: hardware
GPU: Titan, Titan Black, Titan X, K20, K40, K80.
- Region Proposal Network (RPN)
- 2GB GPU memory for ZF net
- 5GB GPU memory for VGG-16 net
- Ojbect Detection Network (Fast R-CNN)
- 3GB GPU memory for ZF net
- 8GB GPU memory for VGG-16 net
but I think these requirements are only for training proccess, in testing process (script_faster_rcnn_demo.m), if you set opts.use_gpu = false, the /functions/nms/nms.m file will not use gpu, thus affects boxes_filter's speed.
As for proposal_im_detect、fast_rcnn_conv_feat_detect、fast_rcnn_im_detect, I didn't find anywhere to use gpu, maybe by caffe.set_mode_gpu().
from faster_rcnn.
@Anylee2142 You should open /experiments/+Dataset/private/voc2007_devkit.m and voc2012_devkit.m, and set your own VOC dataset devkit directory.
Then run script_faster_rcnn_VOC0712_ZF or script_faster_rcnn_VOC2007_ZF,
If the same errors still appear, please set break point at the line (line 46, imdb_from_voc.m ):
imdb.image_ids = textread(sprintf(VOCopts.imgsetpath, image_set), '%s');
to see VOCopts.imgsetpath and check your VOC devkit dirs
from faster_rcnn.
@Cogito2012
Before checking your reply,
i had simply renamed the folder i downloaded as what path indicate:= VOCdevkit2007
and put VOC2007 folder into VOCdevkit2007 because imdb.img_dir is
D:/MATLAB2013/faster_rcnn-master/datasets/VOCdevkit2007/VOC2007/JPEGImages
after doing this, it seemed to run but
while MATLAB console displayed some working message, same error appeared, after this,
i'm still on the same point. (error running train, it doesn't even start)
VOCopts.imgsetpath's dir is
'D:/MATLAB2013/faster_rcnn-master/datasets/VOCdevkit2007/VOC2007/ImageSets/Main/%s.txt'
many files are in Main folder.. i don't know what's wrong
this is information about VOCopts
PATH.zip
from faster_rcnn.
@Anylee2142 I don't actually know whether your local VOC dataset directory is right, I put my VOC2007 at disk E:, because the path in matlab is not permitted too long.
And I suggest that not rename the VOCdevkit, such as
D:/MATLAB2013/faster_rcnn-master/datasets/VOCdevkit/VOC2007/JPEGImages
from faster_rcnn.
@Cogito2012
hmm.. i got folder's name back as VOCdevkit but nothing changed.
May i ask that you tell me your local configuration of datasets, VOC2007 and VOCdevkit?
maybe following your configuration works for me.
from faster_rcnn.
@Anylee2142
VOC2007 datasets in my disk E:/PASCAL VOC2007/VOCdevkit/
|-- VOCdevkit/
|-- local/
|-- results/
|-- VOC2007/
|-- Annotations/
*.xml
|-- ImageSets/
|-- Action/
|-- Layout/
|-- Main/
*.txt (84 text files:train+val+trainval+test)
|-- Segmentation/
|-- JPEGImages/
*.jpg
|-- SegmentationClass/
|-- SegmentationObject/
|-- VOCcode/
*.m
*.m
And in my /experiments/+Dataset/private/voc2007_devkit.m file:
path = 'E:/PASCAL VOC2007/VOCdevkit';
from faster_rcnn.
@Cogito2012
I'm grateful for your good.
My VOCdevkit/VOC2007/ImageSets/Main has only 63 txt files.
they have name like train+val+trainval+test but lesser than 84.
Is the reason why MATLAB displays Error using textread because of missing files?
If i may ask, Could you upload your Imagesets zip folder please?
from faster_rcnn.
@Anylee2142
The 84 files are named as:
class_train.txt (20 files)
train.txt
class_val.txt (20 files)
val.txt
class_trainval.txt (20 files)
trainval.txt
class_test.txt (20 files)
test.txt
In the faster-rcnn training, only the trainval.txt will be used,if you didn't download test datasets, there will be 63 files, but will not affect the training.
Here is my ImageSets
from faster_rcnn.
@Cogito2012
i really don't know what's wrong with my configuration..
why MATLAB can't find trainval.txt .. it's obviously there.
anyway, my testing demo problem is solved, i'm closing this issue.
thank you Cogito2012
from faster_rcnn.
@Anylee2142 and @Cogito2012 : I could run script_faster_rcnn_VOC2007_ZF without any problem with the original VOC2007 Dataset. Now I use my own dataset and I changed all path variables to this new dataset which has exactly the same architecture as VOC2007. I cleared all cachefiles and I get exactly this error.
Only differences: I am more interested in the RPN and hence I have only 2 object classes, no poses and no parts. Do you know if this error can occur if you have wrong prototxt files or so? I had already a problem to create the roidb, where I set my source path manually.
from faster_rcnn.
Related Issues (20)
- demo Check failed HOT 3
- No detection on "training image" with a trained net
- GPU memory is not enough(2g),how to deal with it
- A compiling problem with Win10+VS2017+MATLAB R2016b+CUDA9.1: undefined reference
- the error when training with VOC2007 HOT 2
- Error when compiling nms_gpu_mex
- Error '_MSC_VER': value '1600' doesn't match value '1800' in nms_gpu_mex.o running faster_rcnn_build.m file HOT 1
- training faster rcnn on ROI only ?
- Fast RCNN loss doesn't change when trained with my own dateset. Not fine tuned on pretrained model
- error at hNet = caffe_('get_net', model_file, phase_name); HOT 1
- runtime question
- Some problems when i train VOC2007 database HOT 1
- I just finished demo and try to train with pascal_voc dataset. (1) If I command /py-faster-rcnn/experiments/scripts/faster_rcnn_alt_opt.sh 0 ZF pascal_voc ./faster_rcnn_alt_opt.sh: line 46: ./tools/train_faster_rcnn_alt_opt.py: No such file or directory above error message is shown up. HOT 1
- error when run faster rcnn in (stage one of fast rcnn ) win10, GPU 1080 TI, Matlab 2018a ; i RUN demo without error ; any solve ? HOT 1
- Is there validatoin set in faster_rcnn ?
- Concatenate a fixed value to fc7 features
- how can i train faster R-CNN using a new customised network model?
- Failed when running 'faster_rcnn_build.m' HOT 1
- What is the mean image?
- Loss is not changing
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from faster_rcnn.