Forums - YOLO v3 pb convert dlc error

2 posts / 0 new
or Register
Last post
YOLO v3 pb convert dlc error
Sunny
Join Date: 11 Dec 18
Posts: 1
Posted: Thu, 2019-05-09 22:42
Top

Hi, I'm trying to convert yolo v3 tensorflow model to dlc file. 

first, I have followed this link to convert yolov3 darknet model to tensorflow :https://github.com/jinyu121/DW2TF

second, using tensorflow tools i made the frozen_graph_yolov3.pb file. 

and then, I am trying to covert this pb file to DLC file using this command :

snpe-tensorflow-to-dlc --graph  frozen_yolov3.pb -i yolov3/net1 1,608,608,3 --out_node yolov3/convolutional59/BiasAdd --out_node yolov3/convolutional67/BiasAdd --out_node yolov3/convolutional75/BiasAdd --dlc yolov3.dlc --verbose 

I am getting following error:

. . .

2019-05-09 19:07:18,561 - 89 - DEBUG - DEBUG_TF_OP_NAME_TYPE_PRINT:     Operation(yolov3/convolutional75/kernel) [Const])
2019-05-09 19:07:18,561 - 89 - DEBUG - DEBUG_TF_OP_NAME_TYPE_PRINT:     Operation(yolov3/convolutional75/bias) [Const])
2019-05-09 19:07:18,561 - 89 - DEBUG - DEBUG_TF_OP_NAME_TYPE_PRINT:     Operation(yolov3/convolutional75/Conv2D) [Conv2D])
2019-05-09 19:07:18,561 - 89 - DEBUG - DEBUG_TF_OP_NAME_TYPE_PRINT:     Operation(yolov3/convolutional75/BiasAdd) [BiasAdd])
2019-05-09 19:07:18,561 - 87 - DEBUG - DEBUG_TF_SCOPE_PRINT: Scope(yolov3/convolutional75/kernel)
2019-05-09 19:07:18,561 - 89 - DEBUG - DEBUG_TF_OP_NAME_TYPE_PRINT:     Operation(yolov3/convolutional75/kernel/read) [Identity])
2019-05-09 19:07:18,561 - 87 - DEBUG - DEBUG_TF_SCOPE_PRINT: Scope(yolov3/convolutional75/bias)
2019-05-09 19:07:18,561 - 89 - DEBUG - DEBUG_TF_OP_NAME_TYPE_PRINT:     Operation(yolov3/convolutional75/bias/read) [Identity])
2019-05-09 19:07:22,086 - 393 - WARNING - ERROR_TF_FALLBACK_TO_ONDEMAND_EVALUATION: Unable to resolve operation output shapes in single pass. Using on-demand evaluation!
2019-05-09 19:07:22,087 - 290 - INFO - INFO_ALL_BUILDING_NETWORK: 
==============================================================
Building Network
==============================================================
2019-05-09 19:07:22,091 - 313 - INFO - INFO_TF_BUILDING_INPUT_LAYER: Building layer (INPUT) with node: yolov3/net1, shape [1, 608, 608, 3]
2019-05-09 19:11:51,323 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: 
. . . 
2019-05-09 19:16:30,067 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMul) with nodes: [u'yolov3/convolutional60/Activation/mul']
2019-05-09 19:16:30,075 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMax) with nodes: [u'yolov3/convolutional60/Activation']
2019-05-09 19:16:30,077 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ResizeNearestNeighbor) with nodes: [u'yolov3/upsample1']
/snpe-1.19.2/lib/python/converters/tensorflow/layers/resize.py:109: RuntimeWarning: error_code=902; error_message=Layer parameter value is invalid in DSP. Layer yolov3/upsample1:0: Only supports BiLinear resize mode. ; error_component=DSP Runtime; line_no=525; thread_id=140171445737216
  align_corners=descriptor.align_corners)
2019-05-09 19:16:30,097 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Concatenation) with nodes: [u'yolov3/route2']
2019-05-09 19:16:30,186 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: [u'yolov3/convolutional61/Conv2D', u'yolov3/convolutional61/BatchNorm/FusedBatchNorm']
2019-05-09 19:16:42,665 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMul) with nodes: [u'yolov3/convolutional61/Activation/mul']
2019-05-09 19:16:42,670 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMax) with nodes: [u'yolov3/convolutional61/Activation']
2019-05-09 19:16:42,671 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: [u'yolov3/convolutional62/Conv2D', u'yolov3/convolutional62/BatchNorm/FusedBatchNorm']
2019-05-09 19:16:55,908 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMul) with nodes: [u'yolov3/convolutional62/Activation/mul']
2019-05-09 19:16:55,912 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMax) with nodes: [u'yolov3/convolutional62/Activation']
2019-05-09 19:16:55,914 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: [u'yolov3/convolutional63/Conv2D', u'yolov3/convolutional63/BatchNorm/FusedBatchNorm']
2019-05-09 19:17:09,074 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMul) with nodes: [u'yolov3/convolutional63/Activation/mul']
2019-05-09 19:17:09,079 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMax) with nodes: [u'yolov3/convolutional63/Activation']
2019-05-09 19:17:09,080 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: [u'yolov3/convolutional64/Conv2D', u'yolov3/convolutional64/BatchNorm/FusedBatchNorm']
2019-05-09 19:17:22,238 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMul) with nodes: [u'yolov3/convolutional64/Activation/mul']
2019-05-09 19:17:22,243 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMax) with nodes: [u'yolov3/convolutional64/Activation']
2019-05-09 19:17:22,244 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: [u'yolov3/convolutional65/Conv2D', u'yolov3/convolutional65/BatchNorm/FusedBatchNorm']
2019-05-09 19:17:35,810 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMul) with nodes: [u'yolov3/convolutional65/Activation/mul']
2019-05-09 19:17:35,820 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMax) with nodes: [u'yolov3/convolutional65/Activation', u'yolov3/route3']
2019-05-09 19:17:35,824 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: [u'yolov3/convolutional68/Conv2D', u'yolov3/convolutional68/BatchNorm/FusedBatchNorm']
2019-05-09 19:17:50,640 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (ElementWiseMul) with nodes: [u'yolov3/convolutional68/Activation/mul']
2019-05-09 19:17:50,645 - 471 - INFO - INFO_ALL_BUILDING_LAYER_W_NODES: Building layer (Convolution) with nodes: [u'yolov3/convolutional58/Conv2D', u'yolov3/convolutional58/BatchNorm/FusedBatchNorm']
2019-05-09 19:17:50,688 - 106 - ERROR - Conversion failed: ERROR_TF_INPUT_OPERATION_NOT_FOUND: Input operation not found for yolov3/convolutional58/Conv2D
 
 
Any help will be appreciated 
 
Thanks 
 

 

  • Up0
  • Down0
Re: YOLO v3 pb convert dlc error
jhchris713
Join Date: 25 Jul 19
Posts: 2
Posted: Mon, 2019-08-05 01:13
Top

Hi,

I am having a trouble on using snpe sdk to convert yolov3.

Did you had any progress after this wirte?

If it is, which code did you used for tensorflow version of Yolov3??

Need your help whatever it is.. It works with inceptionV3 which is tutorial of snpe_tensorflow_to_dlc

Thanks,

Chris

  • Up0
  • Down0
or Register

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries (“Qualcomm”). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.