Revision history [back]

I would like to create a .pbtxt from a YoloV3 .pb file trained using TensorFlow, using a python script.

There are already three available python scripts for SSD_Cnn, a Faster_Rcnn and a Mask_Rcnn, but not for YoloV3.

Obviously, YoloV3 inference is implemented in Opencv using Darknet, but for many reasons i would like to use the training results from a YoloV3 Keras-tensorflow implementation.

The question is: is it possible to write a similar script, or there are some fundamental reasons that preclude this possibility?

Is there someone that has tried this solution?

Thanks!

I would like to create a .pbtxt from a YoloV3 .pb file trained using TensorFlow, using a python script.

There are already three available python scripts for SSD_Cnn, a Faster_Rcnn and a Mask_Rcnn, but not for YoloV3.

Obviously, YoloV3 inference is implemented in Opencv using Darknet, but for many reasons i would like to use the training results from a YoloV3 Keras-tensorflow implementation.

The question is: is it possible to write a similar script, or there are some fundamental reasons that preclude this possibility?

Is there someone that has tried this solution?

Thanks!

I would like to create a .pbtxt from a YoloV3 .pb file trained using TensorFlow, using a python script.

There are already three available python scripts for SSD_Cnn, a Faster_Rcnn and a Mask_Rcnn, but not for YoloV3.

Obviously, YoloV3 inference is implemented in Opencv using Darknet, but for many reasons i would like to use the training results from a YoloV3 Keras-tensorflow implementation.

The question is: is it possible to write a similar script, or there are some fundamental reasons that preclude this possibility?

Is there someone that has tried this solution?

Thanks!

I would like to create a .pbtxt from a YoloV3 .pb file trained using TensorFlow, using a python script.

There are already three available python scripts for SSD_Cnn, a Faster_Rcnn and a Mask_Rcnn, but not for YoloV3.

Obviously, YoloV3 inference is implemented in Opencv using Darknet, but for many reasons i would like to use the training results from a YoloV3 Keras-tensorflow implementation.

The question is: is it possible to write a similar script, or there are some fundamental reasons that preclude this possibility?

Is there someone that has tried this solution?

Thanks!