bgshih / seglink Goto Github PK

you have "./exp/sgd/checkpoint" in finetune_ic15.json
but how to get this checkpoint file?,

{
  "cuda_devices": "2,3",
  "log_prefix": "finetune_ic15_",
  "resume": "finetune",
  "finetune_model": "../exp/sgd/checkpoint",
  "train_datasets": "../data/icdar_2015_incidental_train.tf",
  "image_height": 512,
  "image_width": 512,
  "train_batch_size": 32,
  "n_data_threads": 8,
  "max_steps": 10000,
  "optimizer": "sgd",
  "base_lr": 1e-4,
  "lr_policy": "staircase",
  "lr_breakpoints": [2000, 4000, 6000],
  "lr_decay": [1.0, 0.1, 0.01],
  "profiling": 0,
  "brief_summary_period": 20,
  "detailed_summary_period": 200,
  "checkpoint_period": 1000
}

when i run "./manage.py train exp/sgd finetune_ic15" . i got this message:

NotFoundError (see above for traceback): Unsuccessful TensorSliceReader constructor: Failed to find any matching files for ../exp/sgd/checkpoint

how can i solve this problem?

Thanks for the clean and elegant code!
I tried to run training from scratch (use pretrained vgg_16 model on imagenet), the traning process looks weird.

Total Loss

And the corresponding loss for others.

the loss quickly converged to about 10+, and I test the model, but no text boxes is detected, how can I diagnose this?

I just made a mistake :)

Hi, thank you for sharing your code.
However, when I try to do experiment using the code, I found that the /seglink/src/ folder is missing which should contains model_vgg16.py

Then I found this on your github.
But it won't work. Seems like this model_vgg16 is not compatatible with that project.
Could you check and give me some advice? Thank you.

conda env create --name RR-<Team_2863>-stage1 -f environment.yml

help!!!!

Hi.
Is there a way to see the precision of the prediction on the test data in tensorboard?

Also please do elaborate the who process of how to perform demo on custom images.

Hi,
I trained model using synthtext and ICDAR focuse data set.
When I test using ICDAR focuse data set, almost all result rectangles shift to right half to two charactors.
Did I make any error?

Thanks!

hello!
when I run ./manage.py <exp> train,the error shows:

Python error: is a directory, cannot continue

what's wrong with it?

I used three days on the pretrain and here I want to share my experience with someone who are interested in this project.

Because the pretrained model is not provided, you should train it by yourself. You should download the VGG_ILSVRC_16_layers_fc_reduced_deploy.prototxt and VGG_ILSVRC_16_layers_fc_reduced.caffemodel. please pay attention to it, it is ssd vgg16 caffemodle not vgg16 model. Because there is another vgg16.tf file in the github, it is a little misleading. And for the test.py, this file is also for the vgg16 which is classification not detection. I suggest it should be deleted from the source file to avoid the misunderstanding.
The most important part is that you should change the
import model_vgg16 in convert_caffemodel_to_ckpt.py to import model_cnn under the seglink.
Also change the path sys.path.insert(0, os.path.join(parentdir, '../../src'))
to sys.path.insert(0, os.path.join(parentdir, '../../seglink'))

##create network
vgg16 = model_vgg16.Vgg16Model()
to
##create network
vgg16 = model_cnn.SsdVgg16()

But then you will encounter another problem with weight_init_method, you should add tf.app.flags.DEFINE_string("weight_init_method", "xavier", "") in model_cnn.
It will affect the train process, so I suggest to create a new model_cnn_ckpt.cy which is the copy of the model_cnn.py with tf.app.flags.DEFINE_string("weight_init_method", "xavier", "")

Please pay attention to model_scope, I have struggled with it for a longer time (maybe I am not so familiar with tensorflow 233333). All of the model_scope used here is "ssd/vgg16".

That's all what I have remembered so far. I hope these tips are useful for you and bring good luck to my current project :)

Hi,
How do I use seglink?
i.e. detect images and train images with ICDAR 2015 dataset?

Also, can I confirmed if seglink has been correctly compiled in the screenshot below?

Thanks.

'third_party/eigen3/unsupported/Eigen/CXX11/Tensor' file not found

thanks!

where do i find the model.vgg16.py?

hi
thank you for your effort

But you do not have any suggestions to run this depending on icdar2017 instead of icdar13 & icdar15

The old data is no longer available
Thanks again

When I run "./manage.py train exp/sgd pretrain", the error shows:
Traceback (most recent call last):
File "solver.py", line 12, in
import model
File "/home/tengli1107/seglink/seglink/model.py", line 6, in
import ops
File "/home/tengli1107/seglink/seglink/ops.py", line 188, in
oplib = load_oplib(LIB_NAME)
File "/home/tengli1107/seglink/seglink/ops.py", line 22, in load_oplib
oplib = tf.load_op_library(lib_copy_path)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/load_library.py", line 56, in load_op_library
lib_handle = py_tf.TF_LoadLibrary(library_filename)
tensorflow.python.framework.errors_impl.NotFoundError: /tmp/lib956d0286_seglink.so: undefined symbol: _ZTIN10tensorflow8OpKernelE

How can I solve this problem? I'm using ubuntu16.04, tensorflow 1.8, gcc 5.4.
I also tried on ubuntu 14.04, still got the same problem.

Thank you

When I ran python3 ./manage.py build_op, catch the following errors:

/root/workspace/seglink/seglink/cpp/detection_mask_op.cc:6:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
Building CXX object CMakeFiles/seglink.dir/project_polygons_op.cc.o
/root/workspace/seglink/seglink/cpp/encode_groundtruth_op.cc:7:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
/root/workspace/seglink/seglink/cpp/polygons_to_rboxes_op.cc:6:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
/root/workspace/seglink/seglink/cpp/clip_rboxes_op.cc:6:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
/root/workspace/seglink/seglink/cpp/combine_segments_op.cc:9:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
/root/workspace/seglink/seglink/cpp/sample_crop_bbox_op.cc:4:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
/root/workspace/seglink/seglink/cpp/decode_segments_links_op.cc:9:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
/root/workspace/seglink/seglink/cpp/project_polygons_op.cc:5:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory
 #include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
                                                             ^
compilation terminated.
make[2]: *** [CMakeFiles/seglink.dir/polygons_to_rboxes_op.cc.o] Error 1
make[2]: *** Waiting for unfinished jobs....
make[2]: *** [CMakeFiles/seglink.dir/clip_rboxes_op.cc.o] Error 1
make[2]: *** [CMakeFiles/seglink.dir/encode_groundtruth_op.cc.o] Error 1
make[2]: *** [CMakeFiles/seglink.dir/detection_mask_op.cc.o] Error 1
make[2]: *** [CMakeFiles/seglink.dir/project_polygons_op.cc.o] Error 1
make[2]: *** [CMakeFiles/seglink.dir/combine_segments_op.cc.o] Error 1
make[2]: *** [CMakeFiles/seglink.dir/decode_segments_links_op.cc.o] Error 1
make[2]: *** [CMakeFiles/seglink.dir/sample_crop_bbox_op.cc.o] Error 1
make[1]: *** [CMakeFiles/seglink.dir/all] Error 2
make: *** [all] Error 2
cp: cannot stat ‘libseglink.so’: No such file or directory
Building complete

my pip list:

absl-py (0.1.10)
astor (0.6.2)
bleach (1.5.0)
certifi (2018.1.18)
gast (0.2.0)
grpcio (1.10.0)
html5lib (0.9999999)
Markdown (2.6.11)
numpy (1.14.1)
pip (9.0.1)
protobuf (3.5.1)
setuptools (38.5.1)
six (1.11.0)
tensorboard (1.6.0)
tensorflow-gpu (1.6.0)
termcolor (1.1.0)
Werkzeug (0.14.1)
wheel (0.30.0)

is the tensorflow-gpu version wrong?