liuyiwai / simple-iam Goto Github PK

Hi, Liu. Thanks for sharing your work. Now, I meet a problem when training the simple-IAM with multi-gpus. nn.DataParallel works when training the prm classification networks. However, the training process fails when it comes to the iam. Here is my modification on your codes:

self.optimizer_filling = nn.DataParallel(self.optimizer_filling, device_ids=self.Device_ids)
self.optimizer_prm = nn.DataParallel(self.optimizer_prm, device_ids=self.Device_ids)
self.prm_module = nn.DataParallel(peak_response_mapping(self.basebone, **config['model']), device_ids=self.Device_id)
self.filling_module = nn.DataParallel(instance_extent_filling(config), device_ids=self.Device_ids)
self.filling_module.module.load_state_dict(checkpoint['state_dict'], False)
self.prm_module.module.load_state_dict(checkpoint['state_dict'], False)

RuntimeError: Caught RuntimeError in replica 1 on device 1.
Original Traceback (most recent call last):
File "/home/user/anaconda3/envs/CenterMask/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 60, in _worker
output = module(*input, **kwargs)
File "/home/user/anaconda3/envs/CenterMask/lib/python3.7/site-packages/torch/nn/modules/module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "/media/ExtHDD/zzp/simple-IAM-master/iam/modules/instance_extent_filling.py", line 105, in forward
self.channel_num, self.kernel, self.kernel)
RuntimeError: shape '[2, 112, 112, 16, 3, 3]' is invalid for input of size 1806336

Hi! Can you share the code of deal proposal from .mat file to .json with me?

Thanks!

Thanks for your reproduce! Do you have evaluate the performance of IAM？

Hi, nice job!

When I used your code to test on voc2012 test, i met the question following. Could you give me a suggestion?
Thanks very much!

Traceback (most recent call last):
File "main.py", line 89, in
main(args)
File "main.py", line 74, in main
solver.inference(data_loader)
File "/disk4/zwy_data/wsis/simple-IAM/solver.py", line 491, in inference
cmap=plt.cm.get_cmap())
File "/home/harry/anaconda2/envs/siam/lib/python3.6/site-packages/matplotlib/init.py", line 1447, in inner
return func(ax, *map(sanitize_sequence, args), **kwargs)
File "/home/harry/anaconda2/envs/siam/lib/python3.6/site-packages/matplotlib/axes/_axes.py", line 5523, in imshow
im.set_data(X)
File "/home/harry/anaconda2/envs/siam/lib/python3.6/site-packages/matplotlib/image.py", line 698, in set_data
self._A = cbook.safe_masked_invalid(A, copy=True)
File "/home/harry/anaconda2/envs/siam/lib/python3.6/site-packages/matplotlib/cbook/init.py", line 682, in safe_masked_invalid
x = np.array(x, subok=True, copy=copy)
File "/home/harry/anaconda2/envs/siam/lib/python3.6/site-packages/torch/tensor.py", line 449, in array
return self.numpy()
TypeError: can't convert CUDA tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

My environment was following:
certifi 2020.12.5
cycler 0.10.0
Cython 0.29.22
kiwisolver 1.3.1
matplotlib 3.3.4
numpy 1.19.5
opencv-python 3.4.3.18
Pillow 8.1.0
pip 21.0.1
protobuf 3.15.1
pydensecrf 1.0rc3
pyparsing 2.4.7
python-dateutil 2.8.1
PyYAML 5.4.1
scipy 1.5.4
setuptools 49.6.0.post20210108
six 1.15.0
tensorboardX 2.1
torch 1.3.1
torchvision 0.4.2
wheel 0.36.2

Hi, I tried to reproduce the result of IAM, but i failed. Some things I found may help:

1. pre-trained weight of PRM
2. COB proposal, the quality is better than MCG

Hope this information is helpful to you.

Hi! I am reading this paper, and your implementation allows me to better understand the idea of this paper, thanks!

And I have some questions about data set. Is your training data is from PASCAL VOC 2012 segmentation? Only use 1464 images for training, and 1449 images for validation?

Or you are using the "augmented pascal voc 2012 data set", which has 10582 images for training? If so, can you share how to get the augmented pascal voc 2012 data set with me?
I just only find this , but i don't know where to get the raw image and other annotations.

By the way,the link to the Pascal voc 2012 data set is dead, it seems that the official website is broken

Thanks for your help! Have a nice day!