Thanks for the great article, i was able to create my own dataset.
Question : How we can use the train model on other machine.
(trained on windows and want to use the same on Linux machine- Raspberry pi)
I got the cfg file "tiny-yolo-voc-lift1c.cfg",
"weight tiny-yolo-voc.weights" and
load file :" 8625"
i copied the cfg,weights and only checkpoint 8625 files (not all check point files) to raspberry pi (linux bases machine)
on Raspberry pi its not working for me.
sudo ./darknet detect cfg/tiny-yolo-voc-lift1c.cfg 8625 lab2/000005.png
but pre-trained model are woring fine:
sudo ./darknet detect cfg/tiny-yolo-voc.cfg bin/tiny-yolo-voc.weights lab2/dog.png
Could you please give the example for pb and meta part here.
Did you copy all the below 4 files from a check point?
.profile
.meta
.data-00000-of-00001
.index
Thanks Sandeep for looking on my issue, to answer your question :
Yes I copied all 4 files:
tiny-yolo-voc-lift1c-8625.data-00000-of-00001
tiny-yolo-voc-lift1c-8625.index
tiny-yolo-voc-lift1c-8625.meta
tiny-yolo-voc-lift1c-8625.profile
also copy the cfg and weight file.
I want to run this trained model on raspberry pi or any other linux machine.
OK. In that case, do you have the files under darkflow/ckpt? or somewhere else?
yes, files are placed inside ckpt
cfg file to cfg folder
weight to bin folder
Hi,
if I am not mistaken. Your command for creating a *.pb should be the following:
flow --model cfg/tiny-yolo-voc-lift1c.cfg --load -1 --savepb
or in your case if it is the latest ckpt e.g. 8625
flow --model cfg/tiny-yolo-voc-lift1c.cfg --load 8625 --savepb
Does that work for you?
@teamtoshare Thank you, but it doesn't work. Do you have any idea?
Hello,
I am highly interested in the answer of this question, anybody got any answers?
Hello,
I am highly interested in the answer of this question, anybody got any answers?
@AbanoubMamdouh Hi bro, did you fixed it? I have the same problem, expect for you rely
Most helpful comment
Hi,
if I am not mistaken. Your command for creating a *.pb should be the following:
Saving the lastest checkpoint to protobuf (pb) file
flow --model cfg/tiny-yolo-voc-lift1c.cfg --load -1 --savepb
or in your case if it is the latest ckpt e.g. 8625
Saving a specific checkpoint to protobuf (pb) file
flow --model cfg/tiny-yolo-voc-lift1c.cfg --load 8625 --savepb
Does that work for you?