weijun88 / ldf Goto Github PK

作者您好，我前面的train-course里的train.py和test.py, train-fine里的train.py都可以运行成功。只有这个出现了报错。我不知道如何解决。能帮忙看看是哪里出问题了？谢谢~~~~

你好！我最近在复现你的论文，按你的readme中训练过程进行了重新训练，与论文中实验结果有些不同，不知道你在训练时有什么特殊技巧或设置吗。

Hi weijun, congratulations! I'm really interested in your code and can't wait to try it out. So when are you going to release the code? Thank you! The paper was great!

The "iteration decoder" in Figure.2. should be "iteration encoder".

你好，
感谢你做的贡献，我想请问一下你训练的数据集是怎么样的？train.txt里面内容是什么？因为我下DUTS数据集训练的时候，一直给我报错‘TypeError: 'NoneType' object is not callable’。
谢谢

Hi,
I just found your provided saliency maps do not have the same evaluation results on PASCAL-S dataset as reported in the paper. I binarize the PASCAL-S dataset for evaluation.
Did you process the PASCAL-S dataset differently?

I want detect generic object in a picture regradless of classes of the object such as SS(Selective Search), however I have not found any article talk about it in recent 3 year, is there anything wrong?

hello， In your paper, IOU loss cannot be applied when calculating detail and body, because the input of IOU label is in binary form. But I think BASNet is the IOU loss calculated directly using the true saliency probability. How do you understand this? and the target form is 0 or 1 ? or [0,1] saliency probability

As you written in your paper,

In addition, we multiply the
newly generated labels with the original binary image I to
remove the background interference as

But I see no where you do this in the code you published, is there something I missed?
Looking for your reply ASAP.

Traceback (most recent call last):
File "/content/drive/MyDrive/Colab Notebooks/LDF/LDF/train-coarse/apex/apex/parallel/init.py", line 15, in
import syncbn
ModuleNotFoundError: No module named 'syncbn'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "train.py", line 15, in
from apex import apex
File "/content/drive/MyDrive/Colab Notebooks/LDF/LDF/train-coarse/apex/apex/init.py", line 8, in
from . import parallel
File "/content/drive/MyDrive/Colab Notebooks/LDF/LDF/train-coarse/apex/apex/parallel/init.py", line 18, in
from .sync_batchnorm import SyncBatchNorm
File "/content/drive/MyDrive/Colab Notebooks/LDF/LDF/train-coarse/apex/apex/parallel/sync_batchnorm.py", line 5, in
from .sync_batchnorm_kernel import SyncBatchnormFunction
File "/content/drive/MyDrive/Colab Notebooks/LDF/LDF/train-coarse/apex/apex/parallel/sync_batchnorm_kernel.py", line 4, in
from apex.parallel import ReduceOp
ModuleNotFoundError: No module named 'apex.parallel'

What wrong to this apex ?

Your work is valuable and interesting. However, you only provide the results under the metrics of MAE, mean-F measure and E-measure. So I want to learn from your saliency maps. Thanks a lot.

How to get an accuracy of a salient object detection model?

Hello,

I have been trying to get your model up and running. I want to test it with the ECSSD Dataset.

I am running Windows 10 with Cuda 10.1.

I get the following error when running: python test.py

Parameters...
datapath  : ../data/ECSSD
snapshot  : ./out/model-40.pt
mode      : test
Traceback (most recent call last):
  File "test.py", line 49, in <module>
    t.save()
  File "test.py", line 35, in save
    for image, (H, W), name in self.loader:
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\utils\data\dataloader.py", line 352, in __iter__
    return self._get_iterator()
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\utils\data\dataloader.py", line 294, in _get_iterator
    return _MultiProcessingDataLoaderIter(self)
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\utils\data\dataloader.py", line 801, in __init__
    w.start()
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\multiprocessing\process.py", line 121, in start
    self._popen = self._Popen(self)
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\multiprocessing\context.py", line 224, in _Popen
    return _default_context.get_context().Process._Popen(process_obj)
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\multiprocessing\context.py", line 327, in _Popen
    return Popen(process_obj)
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\multiprocessing\popen_spawn_win32.py", line 93, in __init__
    reduction.dump(process_obj, to_child)
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\multiprocessing\reduction.py", line 60, in dump
    ForkingPickler(file, protocol).dump(obj)
TypeError: 'NoneType' object is not callable

C:\Users\Seppi\Desktop\LDF\train-fine>Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\multiprocessing\spawn.py", line 116, in spawn_main
    exitcode = _main(fd, parent_sentinel)
  File "C:\Users\Seppi\AppData\Local\Programs\Python\Python38\lib\multiprocessing\spawn.py", line 126, in _main
    self = reduction.pickle.load(from_parent)
EOFError: Ran out of input

Firstly, thanks for your contributions. I want to know why is the training part divided into "train-coarse" and "train-fine"? And using the result of the "train-coarse" phase to train "train-fine"? Could you please help me understand that? Thanks a lot!