search

nemtiax / MiniTwitter

0 stars 0 forks source link

tof_rgbd_processing-master parameter settings #1

Open cgllv opened 5 years ago

cgllv commented 5 years ago

Hi, nemtiax, recently I am learning about the tof-rgb image alignment and refinment, I read the paper of sylqiu，but when I try to run the program of tof_rgbd_processing-master, I meet some problems. I set the parameters as --dataset2_name nogt, for I think I set the net as test, and I set TestingConfig function inside param self.path='my dataset absolute path', but when I run, it print error: TypeError: Expected list for 'values' argument to 'Pack' Op, not range(0, 512). I'm really confused about the param setting, I see your questions on sylqiu repository. So I hope you can give me some suggestions, thank you very much!

nemtiax commented 5 years ago

Hi Fengweigang, I also haven't been able to get the tof_rgb_processing code to run just yet, but I have been making some progress. I believe I ran into the same issue as you, and what I found was that it's necessary to make sure you have exactly the right version of tensorflow and cuda for the code to run. In particular, installing the old tensorflow 1.2 cleared up the error you describe, if I remember correctly. I'm not sure exactly what all the correct parameter settings are - I'm not sure what exactly dataset2_name is for, but I'll let you know as I figure more out.

On Tue, Nov 5, 2019 at 8:43 PM fengweigang notifications@github.com wrote:

Hi, nemtiax, recently I am learning about the tof-rgb image alignment and refinment, I read the paper of sylqiu，but when I try to run the program of tof_rgbd_processing-master, I meet some problems. I set the parameters as --dataset2_name nogt, for I think I set the net as test, and I set TestingConfig function inside param self.path='my dataset absolute path', but when I run, it print error: TypeError: Expected list for 'values' argument to 'Pack' Op, not range(0, 512). I'm really confused about the param setting, I see your questions on sylqiu repository. So I hope you can give me some suggestions, thank you very much!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/nemtiax/MiniTwitter/issues/1?email_source=notifications&email_token=ABR6WCOMHER2XAUY2DC3CR3QSIOKLA5CNFSM4JJOIWY2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HXC3UZQ, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABR6WCP3PDVVOTN27LSGIF3QSIOKLANCNFSM4JJOIWYQ .

cgllv commented 5 years ago

Hi, nemtiax: I have some the problem proposed yesterday, the reason is python3.5 range() return not a list, I don't know the auther's python version, so I modified by list(range()), but when I run the real data which auther shared several days ago, the net reply the checkpoint is not match, I have set the path of checkpoint, so, I don't know what's wrong, I think maybe the checkpoint file have some problems for I can't load these file to see. the other problem is I don't know how to train the net, I think the dataset2_name is for real dataset that mix real and synthesis image. but the auther didn't share his real dataset, so, have train the net by only synthesis dataset? how to modified the code? may you give me some suggestions?