Any guide for YOLO v2 graph conversion? (by using darkflow with tensorflow)

idata · ‎12-18-2017

The new release v1.11.00 announced to support TinyYolo v2 via Darkflow transformation. Any guide or hint on how to convert the transformed .pb/meta file to graph would be much appreciated.

Many thanks!

PS: I myself tried caffe conversion, it finally worked but suffered from precision loss. It passed mvNCCheck so I guess something wrong with yolo to caffe conversion. The Darkflow conversion works better as the tensorflow model gives testing results as good as Darknet, but I failed to convert the model to NCS graph due to [Error 34]: input contain placeholder. I'm not familiar with TensorFlow so I have no clue about the error.

idata · ‎01-15-2018

Hi,

Have you managed to find out what the issue was please? I would be also interested to know.

Thanks,

Martin Peniak

idata · ‎01-15-2018

Hi,

Any updates on this ??

I Also encountered the same error.

idata · ‎01-19-2018

I got a solution to this in this thread: https://ncsforum.movidius.com/discussion/536/error-34-values-for-input-contain-placeholder