Skip to content

Conversation

@Sundrops
Copy link

@Sundrops Sundrops commented Apr 7, 2018

No description provided.

@EternityZY
Copy link

Use HSV to display the optical flow graph, and then convert the HSV into RGB color map by cv2.cvtColor(hsv, cv2.COLOR_HSV2RGB) for visualization.

@AmazingAng
Copy link

How to train u0? I can not find details in the paper nor the code.

@XDhughie
Copy link

XDhughie commented Oct 8, 2018

Hello:
I'm sorry to disturb you, but have you ever encountered any significant loss when running tvnet? It is exponential loss. How did you solve it
thank u. @ang-13

@liuyuying0829
Copy link

hello ,i have a question about train the tvl1_flow_trainable.py file:
ValueError: Variable tvl1_flow/conv2d/kernel already exists, disallowed. Did you mean to set reuse=True or reuse=tf.AUTO_REUSE in VarScope? Originally defined at:

File "/data1/dzf/tvnet/tvl1_flow_trainable.py", line 20, in grey_scale_image
kernel_initializer=ker_init, use_bias=False, trainable=False,reuse =reuse)
File "/data1/dzf/tvnet/tvl1_flow_trainable.py", line 271, in dual_tvl1_optic_flow_multiscale
grey_x1 = self.grey_scale_image(x1,reuse=False)
File "demo_new.py", line 50, in
max_iterations=iteration)
i find some solve method but failed,can you tell me why and so Thanks.Dear author

@AmazingAng
Copy link

Hello:
I'm sorry to disturb you, but have you ever encountered any significant loss when running tvnet? It is exponential loss. How did you solve it
thank u. @ang-13

no, i dont have this problem.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants