jinyu121 / DW2TF

Darknet Weights to TensorFlow
GNU General Public License v3.0
219 stars 73 forks source link

Converting tiny-yolov3 gives input/output error #32

Open SlavaKeshkov opened 5 years ago

SlavaKeshkov commented 5 years ago

Hello @sjain-stanford . I am trying out the new release that you made for tiny-yolov3 but facing errors during the conversion. I used your code to convert the models:

'python main.py \ --cfg 'data/yolov3-tiny.cfg' \ --weights 'data/yolov3-tiny.weights' \ --output 'data/' \ --prefix 'yolov3-tiny/' \ --gpu 0'

I used original darkflow weights file and the cfg file. However I am getting the error below.

UnknownError (see above for traceback): Failed to rename: data/yolov3-tiny.ckpt.index.tempstate5997668412469460368 to: data/yolov3-tiny.ckpt.index : Access is denied.
; Input/output error

error

As output I get the following files (see screenshot). I am uncertain if the conversion has been completed. Please let me know!

output

albayrakonur commented 4 years ago

did you solve the problem ? I have same one

Robin2091 commented 4 years ago

@albayrakonur Any update? I am also experiencing the same problem.

SlavaKeshkov commented 4 years ago

I switched to another repo for conversion, this one: https://github.com/zzh8829/yolov3-tf2

On Thu, 9 Apr 2020, 17:03 Onur Albayrak, notifications@github.com wrote:

did you solve the problem ? I have same one

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jinyu121/DW2TF/issues/32#issuecomment-611545294, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKWLRBZC6ZBJAXGJNQC6X3LRLXITDANCNFSM4IJ543GA .

Robin2091 commented 4 years ago

@SlavaKeshkov Hi, did you train on darknet with a custom dataset? I also used that repo for conversion(custom dataset with 1 class) but I get worse results.

SlavaKeshkov commented 4 years ago

Yes trained on a custom one. Not sure why there is a difference. I had many classes though, not one. Also look into the threshold value when you convert, it could be causing the issue

On Fri, 10 Apr 2020, 18:35 Robin2091, notifications@github.com wrote:

@SlavaKeshkov https://github.com/SlavaKeshkov Hi, did you train on darknet with a custom dataset? I also used that repo for conversion(custom dataset with 1 class) but I get worse results.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/jinyu121/DW2TF/issues/32#issuecomment-612083053, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKWLRB5AL5QVJZFZOUDLONLRL44DLANCNFSM4IJ543GA .

Robin2091 commented 4 years ago

@SlavaKeshkov Thanks for the quick reply. Are you talking about the IoU threshold or the confidence threshold?

SlavaKeshkov commented 4 years ago

I believe the confidence one

On 10 Apr 2020, at 18:43, Robin2091 notifications@github.com wrote:

@SlavaKeshkov https://github.com/SlavaKeshkov Thanks for the quick reply. Are you talking about the IoU threshold or the confidence threshold?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/jinyu121/DW2TF/issues/32#issuecomment-612086743, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKWLRB24HGY53RCEW6XCXUDRL45DFANCNFSM4IJ543GA.