search

daniilidis-group / EV-FlowNet

Code for the paper "EV-FlowNet: Self-Supervised Optical Flow for Event-based Cameras"
Other
171 stars 44 forks source link

How did you debug this code? #16

Open ChaerinMin opened 2 years ago

ChaerinMin commented 2 years ago

Hi Alex,

Thank you for sharing your great work. I would like to ask exactly what kind of debugger you used for the code. Eager execution seems not allowing Queue which is essential for slim package, and I found Tensorboard debugging doesn't fully support Tensorflow 1.10. Since tf.contrib.slim.learning.train is a single statement and does not show the internal structure of a running graph, I cannot see what is really going on inside the model including tensors and their shapes. I am looking forward your reply.

Thanks

alexzzhu commented 2 years ago

Unfortunately this project was written in early days tensorflow. My best recommendation would be to stick tf.Print commands wherever you need to inspect tensors and running the pipeline.

ChaerinMin commented 2 years ago

I am glad that you answered my question since this is a work from 4 years ago! I really appreciate your kindness. I am now going to try that and hope it would work for me.

alexzzhu commented 2 years ago

Sorry for the late reply! I don't use GitHub in my day to day anymore.

ChaerinMin commented 2 years ago

Then have you got any alternative channel for communications?

alexzzhu commented 2 years ago

I think this is the best avenue for shorter questions, as it leaves a history for other users. If you have longer form questions you can reach me by email.