UFFParser: Parser error: BoxPredictor_0/Reshape: Reshape: -1 dimension specified more than 1 time

Autonomous Machines Jetson & Embedded Systems Jetson Nano
1

When doing the build with ssd_mobilenet_v1 (i just download from TF detection zoon and retrained the model with my own data), i met an error:
[TensorRT] ERROR: UFFParser: Parser error: BoxPredictor_0/Reshape: Reshape: -1 dimension specified more than 1 time

TF version: =1.14.0
Jetpack: 4.2
TensorRT: 5.0

Do you have any ideas? what’s missing here?
Many thanks!

details

amvi@nvidia-nano:~/zhen/tensorrt_demos/ssd$ ./build_engines.sh

for model in ssd_mobilenet_v1_egohands
python3 build_engine.py ssd_mobilenet_v1_egohands
[TensorRT] INFO: Plugin Creator registration succeeded - GridAnchor_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - NMS_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Reorg_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Region_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Clip_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - LReLU_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - PriorBox_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Normalize_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - RPROI_TRT
WARNING:tensorflow:From /usr/lib/python3.6/dist-packages/graphsurgeon/StaticGraph.py:123: FastGFile.init (from tensorflow.python.platform.gfile) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.gfile.GFile.
WARNING: To create TensorRT plugin nodes, please use the create_plugin_node function instead.
UFF Version 0.5.5
=== Automatically deduced input nodes ===
[name: “Input”
op: “Placeholder”
attr {
key: “dtype”
value {
type: DT_FLOAT
}
}
attr {
key: “shape”
value {
shape {
dim {
size: 1
}
dim {
size: 3
}
dim {
size: 300
}
dim {
size: 300
}
}
}
}
]

Using output node NMS
Converting to UFF graph
Warning: No conversion function registered for layer: NMS_TRT yet.
Converting NMS as custom op: NMS_TRT
Warning: No conversion function registered for layer: FlattenConcat_TRT yet.
Converting concat_box_loc as custom op: FlattenConcat_TRT
Warning: No conversion function registered for layer: GridAnchor_TRT yet.
Converting MultipleGridAnchorGenerator as custom op: GridAnchor_TRT
Warning: No conversion function registered for layer: FlattenConcat_TRT yet.
Converting concat_box_conf as custom op: FlattenConcat_TRT
No. nodes: 450
UFF Output written to /home/camvi/zhen/tensorrt_demos/ssd/tmp_v1_egohands.uff
UFF Text Output written to /home/camvi/zhen/tensorrt_demos/ssd/tmp_v1_egohands.pbtxt
[TensorRT] INFO: UFFParser: parsing Input
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_0/weights
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_0/Conv2D
[TensorRT] INFO: UFFParser: Convolution: add Padding Layer to support asymmetric padding
[TensorRT] INFO: UFFParser: Convolution: Left: 0
[TensorRT] INFO: UFFParser: Convolution: Right: 1
[TensorRT] INFO: UFFParser: Convolution: Top: 0
[TensorRT] INFO: UFFParser: Convolution: Bottom: 1
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_0/BatchNorm/gamma
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_0/BatchNorm/beta
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_0/BatchNorm/moving_mean
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_0/BatchNorm/moving_variance
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_0/BatchNorm/FusedBatchNorm
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_0/Relu6
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 9) to 0.0472441
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 11) to 0.0472441
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_1_depthwise/depthwise_weights
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_1_depthwise/depthwise
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_1_depthwise/BatchNorm/gamma
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_1_depthwise/BatchNorm/beta
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_1_depthwise/BatchNorm/moving_mean
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_1_depthwise/BatchNorm/moving_variance
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_1_depthwise/BatchNorm/FusedBatchNorm
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_1_depthwise/Relu6
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 19) to 0.0472441
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 21) to 0.0472441
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_1_pointwise/weights
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_1_pointwise/Conv2D
…
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_10_depthwise/BatchNorm/moving_variance
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_10_depthwise/BatchNorm/FusedBatchNorm
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_10_depthwise/Relu6
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 201) to 0.0472441
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 203) to 0.0472441
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_10_pointwise/weights
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_10_pointwise/Conv2D
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_10_pointwise/BatchNorm/gamma
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_10_pointwise/BatchNorm/beta
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_10_pointwise/BatchNorm/moving_mean
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_10_pointwise/BatchNorm/moving_variance
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_10_pointwise/BatchNorm/FusedBatchNorm
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_10_pointwise/Relu6
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 211) to 0.0472441
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 213) to 0.0472441
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_depthwise/depthwise_weights
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_11_depthwise/depthwise
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_depthwise/BatchNorm/gamma
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_depthwise/BatchNorm/beta
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_depthwise/BatchNorm/moving_mean
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_depthwise/BatchNorm/moving_variance
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_11_depthwise/BatchNorm/FusedBatchNorm
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_11_depthwise/Relu6
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 221) to 0.0472441
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 223) to 0.0472441
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_pointwise/weights
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_11_pointwise/Conv2D
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_pointwise/BatchNorm/gamma
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_pointwise/BatchNorm/beta
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_pointwise/BatchNorm/moving_mean
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/Conv2d_11_pointwise/BatchNorm/moving_variance
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_11_pointwise/BatchNorm/FusedBatchNorm
[TensorRT] INFO: UFFParser: parsing FeatureExtractor/MobilenetV1/MobilenetV1/Conv2d_11_pointwise/Relu6
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 231) to 0.0472441
[TensorRT] INFO: Setting Dynamic Range for (Unnamed ITensor* 233) to 0.0472441
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/BoxEncodingPredictor/weights
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/BoxEncodingPredictor/Conv2D
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/BoxEncodingPredictor/biases
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/BoxEncodingPredictor/BiasAdd
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/Shape
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/strided_slice/stack
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/strided_slice/stack_1
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/strided_slice/stack_2
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/strided_slice
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/Reshape/shape/1
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/Reshape/shape/2
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/Reshape/shape/3
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/Reshape/shape
[TensorRT] INFO: UFFParser: parsing BoxPredictor_0/Reshape
[TensorRT] ERROR: UFFParser: Parser error: BoxPredictor_0/Reshape: Reshape: -1 dimension specified more than 1 time
[TensorRT] ERROR: Network must have at least one output
Traceback (most recent call last):
File “build_engine.py”, line 218, in
main()
File “build_engine.py”, line 212, in main
buf = engine.serialize()
AttributeError: ‘NoneType’ object has no attribute ‘serialize’

2

Hi,

Not sure which tutorial you are using.
But it’s recommended to use this sample which we have verified for ssd_mobilenet_v1 model.
https://github.com/AastaNV/TRT_object_detection

Thanks.

3

Hi AastaLLL,

I followed the tutorial you mentioned. If I used the model directly (ssd_mobilenet_v1), it can work. But when I replace the ssd_mobilenet_v1 with the one (also ssd_mobilenet_v1) that I fine-tuned with my own data, it has following error:

Any ideas?

python3 main.py
[TensorRT] INFO: Plugin Creator registration succeeded - GridAnchor_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - NMS_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Reorg_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Region_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Clip_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - LReLU_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - PriorBox_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - Normalize_TRT
[TensorRT] INFO: Plugin Creator registration succeeded - RPROI_TRT
WARNING:tensorflow:From /usr/lib/python3.6/dist-packages/graphsurgeon/StaticGraph.py:123: FastGFile.init (from tensorflow.python.platform.gfile) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.gfile.GFile.
WARNING: To create TensorRT plugin nodes, please use the create_plugin_node function instead.
Traceback (most recent call last):
File “main.py”, line 30, in
dynamic_graph = model.add_plugin(gs.DynamicGraph(model.path))
File “/home/camvi/zhen/TRT_object_detection/config/model_ssd_mobilenet_v1_coco_2018_01_28.py”, line 79, in add_plugin
graph.find_nodes_by_name(“Input”)[0].input.remove(“image_tensor:0”)
File “/usr/local/lib/python3.6/dist-packages/google/protobuf/internal/containers.py”, line 296, in remove
self._values.remove(elem)
ValueError: list.remove(x): x not in list

4

Some post suggest I need to retrain with the TF1.12.0 and re-do the build. I followed but got another error:

[TensorRT] INFO: For layer (Unnamed Layer* 395) [Shuffle] + BoxPredictor_4/Reshape_1 a higher-precision implementation was chosen than was requested because it resulted in faster network performance
[TensorRT] INFO: Formats and tactics selection completed in 467.253 seconds.
[TensorRT] INFO: After reformat layers: 181 layers
[TensorRT] INFO: Block size 268435456
[TensorRT] INFO: Block size 5837312
[TensorRT] INFO: Block size 2918912
[TensorRT] INFO: Block size 739328
[TensorRT] INFO: Block size 204800
[TensorRT] INFO: Block size 61440
[TensorRT] INFO: Block size 57344
[TensorRT] INFO: Block size 30720
[TensorRT] INFO: Block size 20992
[TensorRT] INFO: Block size 17408
[TensorRT] INFO: Block size 9728
[TensorRT] INFO: Block size 9216
[TensorRT] INFO: Block size 2560
[TensorRT] INFO: Block size 1536
[TensorRT] INFO: Block size 1024
[TensorRT] INFO: Block size 512
[TensorRT] INFO: Total Activation Memory: 278348288
python3: nmsPlugin.cpp:136: virtual void nvinfer1::plugin::DetectionOutput::configureWithFormat(const nvinfer1::Dims*, int, const nvinfer1::Dims*, int, nvinfer1::DataType, nvinfer1::PluginFormat, int): Assertion `numPriors * param.numClasses == inputDims[param.inputOrder[1]].d[0]’ failed.
Aborted (core dumped)

5

Hi,

It looks like some layer names are updated in your customized model.
Could you run the following command for detail and share with us?

$ sudo python3 /usr/lib/python3.6/dist-packages/uff/bin/convert_to_uff.py frozen_inference_graph.pb -l

Thanks.

6

Hi AastaLLL,

The result is in the attached .txt file.
own_model.txt (172 KB)

7

Hi AastaLLL,

I notice that the model is finetuned in python2 environment and my Nano is in python3 environment. Is this an issue?

8

Hi, @xiamenhai,have you solved the problem “ERROR: UFFParser: Parser error: xxxx/Reshape: Reshape: -1 dimension specified more than 1 time”.Looking forward to your reply

9

Hi,

Sorry for the delay.

It looks like the output layer name of your model is ‘add_6’.
Please help to update it into this line and try it again.
https://github.com/AastaNV/TRT_object_detection/blob/master/config/model_ssd_mobilenet_v1_coco_2018_01_28.py#L5

Thanks.