Need help in using jetson-inference ssd v2 with deepstream

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

I followed [Preformatted text](https://elinux.org/index.php?title=Jetson/L4T/TRT_Customized_Example#Custom_Parser_for_SSD-MobileNet_Trained_by_Jetson-inference) guide to use deepstream jetson-inference custom model.

My device is JP 4.5 with DS 5.1 on Xavier NX

Sharing my config:

[property]
gpu-id=0
net-scale-factor=0.0078431372
offsets=127.5;127.5;127.5
model-color-format=0
model-engine-file=ssd-mobilenet.onnx.1.1.7103.GPU.FP16.engine
onnx-file=ssd-mobilenet.onnx
labelfile-path=labels.txt
infer-dims=3;300;300
uff-input-order=0
uff-input-blob-name=Input

0=FP32, 1=INT8, 2=FP16 mode[property]

gpu-id=0
net-scale-factor=0.0078431372
offsets=127.5;127.5;127.5
model-color-format=0
model-engine-file=ssd-mobilenet.onnx.1.1.7103.GPU.FP16.engine
onnx-file=ssd-mobilenet.onnx
labelfile-path=labels.txt
infer-dims=3;300;300
uff-input-order=0
uff-input-blob-name=Input

0=FP32, 1=INT8, 2=FP16 mode

network-mode=2
num-detected-classes=2
interval=0
gie-unique-id=1
is-classifier=0
output-blob-names=MarkOutput_0
parse-bbox-func-name=NvDsInferParseCustomSSD
custom-lib-path=nvdsinfer_custom_impl_ssd/libnvdsinfer_custom_impl_ssd.so
#scaling-filter=0
#scaling-compute-hw=0

[class-attrs-all]
threshold=0.5
roi-top-offset=0
roi-bottom-offset=0
detected-min-w=0
detected-min-h=0
detected-max-w=0
detected-max-h=0

Per class configuration

#[class-attrs-2]
#threshold=0.6
#roi-top-offset=20
#roi-bottom-offset=10
#detected-min-w=40
#detected-min-h=40
#detected-max-w=400
#detected-max-h=800

network-mode=2
num-detected-classes=2
interval=0
gie-unique-id=1
is-classifier=0
output-blob-names=MarkOutput_0
parse-bbox-func-name=NvDsInferParseCustomSSD
custom-lib-path=nvdsinfer_custom_impl_ssd/libnvdsinfer_custom_impl_ssd.so
#scaling-filter=0
#scaling-compute-hw=0

[class-attrs-all]
threshold=0.5
roi-top-offset=0
roi-bottom-offset=0
detected-min-w=0
detected-min-h=0
detected-max-w=0
detected-max-h=0

Per class configuration

#[class-attrs-2]
#threshold=0.6
#roi-top-offset=20
#roi-bottom-offset=10
#detected-min-w=40
#detected-min-h=40
#detected-max-w=400
#detected-max-h=800

My error after running the script:

INFO: [TRT]: Some tactics do not have sufficient workspace memory to run. Increasing workspace size may increase performance, please check verbose output.

INFO: [TRT]: Detected 1 inputs and 4 output network tensors.
0:16:01.296449381 18313 0x36368a00 INFO nvinfer gstnvinfer.cpp:619:gst_nvinfer_logger: NvDsInferContext[UID 1]: Info from NvDsInferContextImpl::buildModel() <nvdsinfer_context_impl.cpp:1749> [UID = 1]: serialize cuda engine to file: /opt/nvidia/deepstream/deepstream-5.1/sources/objectDetector_SSD/ssd-mobilenet.onnx_b1_gpu0_fp16.engine successfully
INFO: [Implicit Engine Info]: layers num: 3
0 INPUT kFLOAT input_0 3x300x300
1 OUTPUT kFLOAT scores 3000x3
2 OUTPUT kFLOAT boxes 3000x4

ERROR: [TRT]: INVALID_ARGUMENT: Cannot find binding of given name: MarkOutput_0
0:16:01.304274698 18313 0x36368a00 WARN nvinfer gstnvinfer.cpp:616:gst_nvinfer_logger: NvDsInferContext[UID 1]: Warning from NvDsInferContextImpl::checkBackendParams() <nvdsinfer_context_impl.cpp:1670> [UID = 1]: Could not find output layer ‘MarkOutput_0’ in engine
0:16:01.329662584 18313 0x36368a00 INFO nvinfer gstnvinfer_impl.cpp:313:notifyLoadModelStatus: [UID 1]: Load new model:/opt/nvidia/deepstream/deepstream-5.1/sources/objectDetector_SSD/config_infer_primary_ssd.txt sucessfully
Decodebin child added: source

Decodebin child added: decodebin0

Decodebin child added: decodebin1

Decodebin child added: rtppcmudepay0

Decodebin child added: rtph265depay0

Decodebin child added: mulawdec0

In cb_newpad

Decodebin child added: h265parse0

Decodebin child added: capsfilter0

Decodebin child added: nvv4l2decoder0

Opening in BLOCKING MODE
Opening in BLOCKING MODE
NvMMLiteOpen : Block : BlockType = 279
NVMEDIA: Reading vendor.tegra.display-size : status: 6
NvMMLiteBlockCreate : Block : BlockType = 279
In cb_newpad

Could not find NMS layer buffer while parsing
0:16:03.730012257 18313 0x3642fe80 ERROR nvinfer gstnvinfer.cpp:613:gst_nvinfer_logger: NvDsInferContext[UID 1]: Error in NvDsInferContextImpl::fillDetectionOutput() <nvdsinfer_context_impl_output_parsing.cpp:724> [UID = 1]: Failed to parse bboxes using custom parse function
Segmentation fault

3

Hi,

Since we have new JetPack 4.6.1 and Deepstream 6.0.1 releases.
It’s recommended to upgrade your environment to the latest first.

We are going to check this issue internally.
Will share more information with you later.

Thanks.

4

Hi,

The customization from the eLinux page is for uff model.
Would you mind sharing your ONNX model so we can check this for you?

Thanks.

5

Hi @AastaLLL - I have a system with JP4.6.1, is it simple with the new JP?
How to execute it?

6

Hi @AastaLLL - I cannot share it with you here on the forum. If there is a way to email or send in DM, I will be fine doing that.

Thanks

7

Hi,

YES. The latest software is JetPack 4.6.1 with Deepstream 6.0.1.
For the model, would you mind sharing it with a private message?

Thanks.

8

Thanks.

We got the model from the private message.
Will update more information here later.

1 Like
9

Hi,

We try your ONNX on JetPack 4.6.1 with Deepstream 6.0.1.
It can work well with a pre-generated engine file from trtexec.

Would you mind also giving it a try?

1. Apply patch from eLinux

$ cd /opt/nvidia/deepstream/deepstream-6.0/sources/objectDetector_SSD/
$ sudo wget https://raw.githubusercontent.com/AastaNV/eLinux_data/main/deepstream/ssd-jetson_inference/ssd-jetson_inference.patch
$ sudo git apply ssd-jetson_inference.patch
$ sudo CUDA_VER=10.2 make -C nvdsinfer_custom_impl_ssd/

2. Generate engine file

$ sudo /usr/src/tensorrt/bin/trtexec --onnx=ssd-mobilenet.onnx --saveEngine=ssd-mobilenet.trt

3. Update configure with the engine file

In config_infer_primary_ssd.txt

...
model-engine-file=ssd-mobilenet.trt

Thanks.

11

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.