error in converting openvino to tflite

[zeckireck]

Issue Type

Others

OS

Ubuntu

OS architecture

x86_64

Programming Language

Python

Framework

OpenVINO, TensorFlowLite

Download URL for ONNX / OpenVINO IR

openvino.zip

Convert Script

--model_path ./openvino/FP32/pphuman_new.xml --output_no_quant_float32_tflite --string_formulas_for_normalization 'data / 255'

Description

i want to convert pphuman_v2.onnx to tflite,but get the below errors, pls help me!

Relevant Log Output

ERROR: {{function_node __wrapped__GatherV2_device_/job:localhost/replica:0/task:0/device:CPU:0}} indices[0] = -1 is not in [0, 4) [Op:GatherV2]
ERROR: model_path  : ./openvino/FP32/pphuman_new.xml
ERROR: weights_path: ./openvino/FP32/pphuman_new.bin
ERROR: layer_id    : 924
ERROR: input_layer0 layer_id=921: tf.Tensor([  1 192 192   2], shape=(4,), dtype=int64)
ERROR: input_layer1 layer_id=922: Const(ndarray).shape  (1,)
array([-1])
ERROR: input_layer2 layer_id=923: Const(ndarray).shape  (1,)
array([0])
ERROR: The trace log is below.
Traceback (most recent call last):
  File "/usr/local/bin/openvino2tensorflow", line 3759, in convert
    tf_layers_dict[layer_id] = tf.gather(
  File "/usr/local/lib/python3.8/dist-packages/tensorflow/python/util/traceback_utils.py", line 153, in error_handler
    raise e.with_traceback(filtered_tb) from None
  File "/usr/local/lib/python3.8/dist-packages/tensorflow/python/eager/execute.py", line 54, in quick_execute
    tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
tensorflow.python.framework.errors_impl.InvalidArgumentError: {{function_node __wrapped__GatherV2_device_/job:localhost/replica:0/task:0/device:CPU:0}} indices[0] = -1 is not in [0, 4) [Op:GatherV2]
ERROR: Please refer to 6-7 in the README first. https://github.com/PINTO0309/openvino2tensorflow

Source code for simple inference testing code

No response

[zeckireck]

pphumanv2.zip this is pphumanv2.onnx

[PINTO0309]

First, follow the instructions in the error message and then ask the question. I am very fed up with all the same questions being posted.

ERROR: Please refer to 6-7 in the README first. https://github.com/PINTO0309/openvino2tensorflow

[zeckireck]

sorry, i have tried to follow the 6-7 instructions , i use this weight_replacement_config_sample.zip to fix the error. but i use the converted tflite model to test the picture, its output seems wrong. i dont know why. pls help me!

[zeckireck]

OpenVINO output @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ elapsed time: 4.8351287841796875ms shape: (1, 2, 192, 192) array([[[[0.9663402 , 0.9663402 , 0.9663843 , ..., 0.97900313, 0.97851515, 0.97851515], [0.9663402 , 0.9663402 , 0.9663843 , ..., 0.97900313, 0.97851515, 0.97851515], [0.9661546 , 0.9661546 , 0.96603453, ..., 0.97858983, 0.9781536 , 0.9781536 ], ..., [0.9730858 , 0.9730858 , 0.9736771 , ..., 0.942102 , 0.9384238 , 0.9384238 ], [0.9740615 , 0.9740615 , 0.97480416, ..., 0.94147056, 0.93764687, 0.93764687], [0.9740615 , 0.9740615 , 0.97480416, ..., 0.94147056, 0.93764687, 0.93764687]],

    [[0.03365977, 0.03365977, 0.0336157 , ..., 0.02099684,
      0.02148488, 0.02148488],
     [0.03365977, 0.03365977, 0.0336157 , ..., 0.02099684,
      0.02148488, 0.02148488],
     [0.03384542, 0.03384542, 0.03396554, ..., 0.02141015,
      0.02184643, 0.02184643],
     ...,
     [0.02691418, 0.02691418, 0.02632285, ..., 0.05789805,
      0.06157618, 0.06157618],
     [0.02593851, 0.02593851, 0.02519586, ..., 0.05852949,
      0.0623531 , 0.0623531 ],
     [0.02593851, 0.02593851, 0.02519586, ..., 0.05852949,
      0.0623531 , 0.0623531 ]]]], dtype=float32)

tflite output @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ elapsed time: 18.19753646850586ms shape: (1, 192, 192, 2) array([[[[1., 1.], [1., 1.], [1., 1.], ..., [1., 1.], [1., 1.], [1., 1.]],

    [[1., 1.],
     [1., 1.],
     [1., 1.],
     ...,
     [1., 1.],
     [1., 1.],
     [1., 1.]],

    [[1., 1.],
     [1., 1.],
     [1., 1.],
     ...,
     [1., 1.],
     [1., 1.],
     [1., 1.]],

    ...,

    [[1., 1.],
     [1., 1.],
     [1., 1.],
     ...,
     [1., 1.],
     [1., 1.],
     [1., 1.]],

    [[1., 1.],
     [1., 1.],
     [1., 1.],
     ...,
     [1., 1.],
     [1., 1.],
     [1., 1.]],

    [[1., 1.],
     [1., 1.],
     [1., 1.],
     ...,
     [1., 1.],
     [1., 1.],
     [1., 1.]]]], dtype=float32)

[zeckireck]