lzgmatrix / bop19_cdpn_2019iccv Goto Github PK

Hi! Thank you for making your great work available!

I was wondering if you could also make your trained models available in another platform besides Baidu. It is quite difficult to download files from there.

Thank you in advance!

Current Dataset: lmo
loading annotations into memory...
Done (t=0.00s)
creating index...
index created!
[ ] 0/95, elapsed: 0s, ETA:Traceback (most recent call last):
File "inference.py", line 170, in
main()
File "inference.py", line 162, in main
result_dict = single_gpu_test(model, data_loader)
File "inference.py", line 61, in single_gpu_test
result = model(return_loss=False, rescale=True, **data)
File "/home/zhuyu/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/home/zhuyu/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 150, in forward
return self.module(*inputs[0], **kwargs[0])
File "/home/zhuyu/anaconda3/envs/open-mmlab/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/home/D/lxj/PoseEst/mmdetection-1.2.0/mmdet/core/fp16/decorators.py", line 49, in new_func
return old_func(*args, **kwargs)
TypeError: forward() missing 1 required positional argument: 'img_metas'

When I refer to test.py to code train.py , I find I can't make it. My loss.backward() can't work!

            output_coor_x_ = output_coor_x_.squeeze()
            output_coor_y_ = output_coor_y_.squeeze()
            output_coor_z_ = output_coor_z_.squeeze()

           ####

            output_coor_ = torch.stack([torch.argmax(output_coor_x_, axis=0),
                                     torch.argmax(output_coor_y_, axis=0),
                                     torch.argmax(output_coor_z_, axis=0)], axis=2)
            output_coor_[output_coor_ == cfg.network.coor_bin] = 0
            output_coor_ = 2.0 * output_coor_.float() / (63.0-1.0) - 1.0      # [-1,1]

I try to use variable with before '####', it work; but when I use variable after '####', it not work and output as follow:

RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn

I am sure that parameters's requires_grad=True in the model.

Maybe I shouldn't use torch.argmax() ?

Could help me solve this problem or tell me how you trian it?

Thank you!

Hello, Mr li, I can`t find the translation head in resnet18, and T_vercor is coming from PnP, Why isn't the highlight of your paper in the code?

Thanks for your code first. I encountered the error below when trying to run the code.

Traceback (most recent call last):
File "main.py", line 132, in
main()
File "main.py", line 125, in main
worker_init_fn=_worker_init_fn()
File "main.py", line 118, in _worker_init_fn
np.random.seed(np_seed)
File "mtrand.pyx", line 244, in numpy.random.mtrand.RandomState.seed
File "_mt19937.pyx", line 166, in numpy.random._mt19937.MT19937._legacy_seeding
File "_mt19937.pyx", line 180, in numpy.random._mt19937.MT19937._legacy_seeding
ValueError: Seed must be between 0 and 2**32 - 1

Then I outputted the torch_seed and np_seed finding that torch_seed = 1648085986 and np_seed = -1.
I think it's because the line116 np_seed = torch_seed // 2**32 - 1 in the function on_worker_init_fn() of main.py
Maybe it can be solved using np_seed = torch_seed // (2**32 - 1) or np_seed = torch_seed % 2**32?

Is detection/annotations/test_lmo.json incomplete?

Hello, Mr Li. Sorry to be a bother. I want to ask why the output channel of the network is 2+195=197? I carefully read the code, If I don't get it wrong, I know the first 2 channels is confidence map and the second 195=65*3 channels is coordinate map that describe x y z map respectively. Why are there 65 channels for x y z map respectively?
At test time, I find that you get the final coordinate-confidence map using argmax like this:
output_coor_ = np.stack([np.argmax(output_coor_x_, axis=0), np.argmax(output_coor_y_, axis=0), np.argmax(output_coor_z_, axis=0)], axis=2)

output_conf_ = np.argmax(output_conf_, axis=0)
I observed that the file 'cfg.yaml' contain the 'coor_bin=64', how to understand this set? The final question is how to set the ground truth when training the network? Looking forward to your reply.