prunetruong / densematching Goto Github PK

Thanks for your great works.
I fail to find

'train_split': 'train_scenes_MegaDepth.txt',
'train_debug_split': 'train_debug_scenes_MegaDepth.txt',
'val_split': 'validation_scenes_MegaDepth.txt',

these files. Could you release more details to preprocess MegaDepth Dataset?

Hi,I want to use WarpC-RANSAC-Flow to warp my own pair images. Can you share the corresponding model and test code ? Thanks!

Hi !
Thanks for sharing such wonderful works since GLAMpoints. All the papers and code greatly helped my research, which I am highly appreciated. I'm looking forward to see the next paper you will release in the near future (perhaps for ECCV)..

I have a question about the newly accepted paper by ICCV'21. Currently, the code seems to be not released yet. If it doesn't bother you, can i please know when will the code be available?

Thanks

Hi, I have some questions about the supervise loss as follows;
ss_loss_o, ss_stats_o = self.objective(estimated_flow_target_prime_to_target_directly, mini_batch['flow_map'], mask=mini_batch['mask'],
the mini_batch['flow_map'] is the ground-truth flow you generate in the online_triplet_creation.py (flow_gt = self.synthetic_flow_generator(mini_batch=mini_batch, training=training, net=net)) and the target image prime is warped by using the flow_gt, so the mini_batch['flow_map'] should be flow_target_to_target_prime_directly, why you caclulate the L1 distance between estimated_flow_target_prime_to_target_directly and mini_batch['flow_map'] rather than the distance between estimated_flow_target_to_target_prime_directly and mini_batch['flow_map'].
Thanks, Looking forward to your reply!

Hi, thanks for your great work.
Would you release the code of PDCNet?

Hi, I'm having some problems with image matching using your project. I used my own dataset, but the matching was poor, especially with a small number of iterations, and the resulting matched images were distorted. I followed the documentation, but the problem persists. I would like to ask for some suggestions to improve the quality of image matching, especially on how to avoid image distortion.

and what's the relationship between the estimated flow miu and the output flow y..?

Traceback (most recent call last): File "test_models.py", line 8, in <module> from model_selection import select_model File "/DenseMatching/model_selection.py", line 1, in <module> from models.GLUNet.GLU_Net import GLUNetModel File "/DenseMatching/models/GLUNet/GLU_Net.py", line 7, in <module> from models.base_matching_net import BaseGLUMultiScaleMatchingNet, set_glunet_parameters File "/DenseMatching/models/base_matching_net.py", line 12, in <module> from third_party.GOCor.GOCor.global_gocor_modules import GlobalGOCorWithFlexibleContextAwareInitializer ModuleNotFoundError: No module named 'third_party'

I meet this error when testing on my own pairs, could you tell me why? Thanks!

python run_training.py PWarpC train_weakly_supervised_PWarpC_SFNet_spair_from_pfpascal. this problem occur when I run the above order.

Hi, Thanks for this wonderful project!
I want to train an optical flow network on my own dataset, but the view point changes in my dataset is small , Can the WarpC loss get a good result in such dataset? I found the dataset used in this project have a large view point changes.

Hi，how much memory storage is required for the dataset MegaDepth? How much memory is required for training the WarpC model on the dataset MegaDepth?

When running the demo_single_pair.py, there report an assert error:

Traceback (most recent call last):
File "D:/PyProjects/DenseMatching-main/demos/demo_single_pair.py", line 103, in
network, estimate_uncertainty = select_model(
File "D:\PyProjects\DenseMatching-main\model_selection.py", line 234, in select_model
network = load_network(network, checkpoint_path=checkpoint_fname)
File "D:\PyProjects\DenseMatching-main\model_selection.py", line 28, in load_network
checkpoint_dict = torch.load(checkpoint_path)
File "E:\ProgramData\anaconda3\envs\pytorch\lib\site-packages\torch\serialization.py", line 592, in load
return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
File "E:\ProgramData\anaconda3\envs\pytorch\lib\site-packages\torch\serialization.py", line 851, in _load
result = unpickler.load()
ModuleNotFoundError: No module named 'admin.local'

network = load_network(network, checkpoint_path=checkpoint_fname)
checkpoint_dict = torch.load(checkpoint_path, map_location='cpu')

I set pre_trained_models path as path_to_pre_trained_models = '../pre_trained_models/PDCNet_plus_megadepth.pth.tar'.
how to fix it?

Hello, thanks for your excellent works and codes. Here, i have an issue for the line 442 in PDCNet.py. I think the codes should be "c_t=c23, c_s=c13 ", which means the source is image1 and target is image2. (Maybe I wrongly understand your code, please correct me (＠_＠;))

Hi, I'm curious about that how to generate the theta for homo/tps/affine. Where should I find the code of the part?

Thanks.

Hi Prune,

Thanks for your great work!

I have noticed that you have some checks for the grid_sample function's behavior change after torch version >= 1.3

These checks look like;
if float(torch.__version__[:3]) >= 1.3:

However, if the torch version is 1.10.x, this check fails (since it compares 1.1 and 1.3 then returns False) and affects the flow values.

I compare the versions with the following lines, and it works properly.

from packaging import version
if version.parse(torch.__version__) >= version.parse("1.3"):

Hi,
Thanks for this wonderful project. I would like to know whether the mask is involved in the loss calculation during training.

Hi,I test the PDCNet and visualize the confidence map with several images.I find the maximum of the confidence is always 0.573 even the query and reference are the same.Does the confidence map mean the uncertaincy of each pixel mapping to [0,1]?

Hi! I'm trying to use the PDC-Net to warp my own pair images, which are shown below.


However, there is a problem said

I don't know how to solve it. I will very appreciate it if you could answer me. Thanks!

1. Is it possible to only get the warped image rather than several images bond together?
2. Is the information of the warped image, say resolution, completely comes from the query image? I mean the reference image doesn't contribute any additional information to the final result.

I think there is a problem with utils_flow/pixel_wise_mapping.py:warp [link] if x, the image to warp, im2, and flo, the dense flow, are different sizes.

flo has shape [H_1, W_1, xy] with x-range (-W_2, W_2) and y-range (-H_2, H_2) and, as noted in the documentation, maps im2 (aka x) with shape [H_2, W_2] back to im1 with shape [H_1, W_1] (via torch.nn.functional.grid_sample). However, here flo is rescaled to [-1, 1] by H_1, W_1, not H_2, W_2 as it should be.
Perhaps this limitation is suggested by the documentation, since it says the shape of x and flo are both (H,W), but that should be made explicit and even ensured with an assert I think.

This affects several other functions: utils_flow/pixel_wise_mapping.py:warp_with_mapping, and unormalise_and_convert_mapping_to_flow, unormalise_flow_or_mapping, unnormalize, & normalize in utils_flow/flow_and_mapping_operations.py. utils_flow/pixel_wise_mapping.py:remap_using_flow_fields works as expected because it does not rescale the flow.

Hei! Great code and organized github! Thanks for your work ;)
I'm using your weights to align images using RANSAC-flow. The only problem that I have is that i'm getting some deformation in the images. I look around your code and in your WarpC paper bud didn't find any detail about the size of the used kernel. In ransac-flow was 7. Did you use the same size?

Thanks!

Hi, I recently trained the PDCNet on Scannet dataset but get a worse result. The flow estimation is regular while the probabilistic map is predicted wrongly. Is it something different between scannet and Megadepth?

Hi,
Thanks for your excellent work. I would like to know if all the experiments in the paper are based on the fixed settings.

Thanks.

Thanks so much for this great work! Do you have an estimated date for the training code to be released?

Thanks!

PruneTruong/GLU-Net#11 (comment)

Thanks,

MegaDepth
We use the reconstructions provided in the D2-Net repo. You can download the undistorted reconstructions and aggregated scene information folder directly here - Google Drive.

However, this link is filed to open,
could you provided it again.

To those, who are trying to run on later versions of cupy: just replace cupy.util.memoize with cupy.memoize