avirambh / scopeflow Goto Github PK
View Code? Open in Web Editor NEWDynamic Scene Scoping for Optical Flow (CVPR 2020)
License: Apache License 2.0
Dynamic Scene Scoping for Optical Flow (CVPR 2020)
License: Apache License 2.0
Dear Authors,
1). I got occlusion maps of all zeros when evaluating using the pretrained model "checkpoints/KITTI_comb/checkpoint_best.ckpt".
2). However, when I overwrite "checkpoints/KITTI_comb/checkpoint_best.ckpt" with the files moving from "checkpoints/Sintel_comb/checkpoint_best.ckpt", and did the same thing in 1), the occlusion map is not zero.
Was the occlusion being trained when creating the checkpoint "checkpoints/KITTI_comb/checkpoint_best.ckpt"?
Thank you very much!
=====
PS.
My environment is
python 3.7.3
pytorch 1.1.0
CUDA 9.0
The correlation package has been installed based on the instruction in the IRR-PWC repository.
Hi!
Thanks for sharing such an excellent work.
I have read the paper and code of ScopeFlow, And I found something confusing problems blow:
1.augmentation.py is used to perform geometry augmentation, and you have set min_rotate and max_rotate to -0.2 and 0.2. But the rotate coefficient will be generate randomly between [0.2, 0.2] such as line 261 or line 791, you set rotate coefficient to [-0.2, 0.2], but when generate random coefficent you use [-min_rotate, max_rotate], and thus rotate coefficient is generated between [0.2,0.2], which means always 0.2.
phi.uniform_(-min_rotate, max_rotate)
min_rotate=-1 * self._args.aug_rotate, max_rotate=self._args.aug_rotate,
aug_rotate: 0.2
2.According to the find_invalid() in file augmentation, when random generated zoom < 1.0, the affine matrix will never be valid, as if zoom<1.0, the corners of transformed img will be out of the border of source img.
Expecting for your replies! Thanks again!
Validate: 0% 0/23 [00:00<?, ?it/s]error in correlation_forward_cuda_kernel: invalid device function
Traceback (most recent call last):
File "evaluate.py", line 41, in <module>
evaluate()
File "evaluate.py", line 37, in evaluate
validation.validate(args)
File "/content/ScopeFlow/lib/validation.py", line 93, in validate
res = train_and_val(args)
File "/content/ScopeFlow/lib/pipeline_wrapper.py", line 87, in train_and_val
validation_augmentation=validation_augmentation)
File "/content/ScopeFlow/lib/runtime.py", line 676, in exec_runtime
augmentation=validation_augmentation).run()
File "/content/ScopeFlow/lib/runtime.py", line 528, in run
loss_dict_per_step, output_dict, batch_size = self._step(example_dict)
File "/content/ScopeFlow/lib/runtime.py", line 451, in _step
loss_dict, output_dict = self._model_and_loss(example_dict)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 477, in __call__
result = self.forward(*input, **kwargs)
File "/content/ScopeFlow/lib/configuration.py", line 85, in forward
output_dict = self._model(example_dict)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 477, in __call__
result = self.forward(*input, **kwargs)
File "/content/ScopeFlow/models/IRR_PWC_V2.py", line 120, in forward
out_corr_relu_f = self._correlate(x1, x2_warp)
File "/content/ScopeFlow/models/IRR_PWC_V2.py", line 275, in _correlate
out_corr = self.correlation(x1, x2_warp)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 477, in __call__
result = self.forward(*input, **kwargs)
File "/content/ScopeFlow/models/correlation_package/correlation.py", line 59, in forward
result = CorrelationFunction(self.pad_size, self.kernel_size, self.max_displacement, self.stride1, self.stride2, self.corr_multiply)(input1, input2)
File "/content/ScopeFlow/models/correlation_package/correlation.py", line 27, in forward
self.pad_size, self.kernel_size, self.max_displacement,self.stride1, self.stride2, self.corr_multiply)
RuntimeError: CUDA call failed (correlation_forward_cuda at correlation_cuda.cc:79)
How do i fix this error?
Thank you for your great work! what are the necessary steps to test your given model on my own datasets?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.