• Hardware Platform (Jetson / GPU) Jetson Nano
• DeepStream Version 5.0
• JetPack Version (valid for Jetson only) 4.4
• TensorRT Version 7.1
• Issue Type( questions, new requirements, bugs) question
- Trained a SSD mobilenet v1 model using custom dataset and convert to onnx model format following instructions here----- https://github.com/dusty-nv/jetson-inference/blob/master/docs/pytorch-ssd.md
- checked if generated onnx model is compatible with deepstream by running /usr/src/tensorrt/bin/trtexec --onnx=/home/jetson/videostreams/samples/models/onx/ssd-mobilenet.onnx
the output i got here is:
&&&& PASSED TensorRT.trtexec # /usr/src/tensorrt/bin/trtexec --onnx=/home/jetson/videostreams/samples/models/onx/ssd-mobilenet.onnx
which implies that the model is compatible to use with deepstrream
- Built custom parser from /opt/nvidia/deepstream/deepstream-5.0/sources/objectDetector_SSD
- modifed the follwing in config file for nvinfer as follows:
model-engine-file=samples/models/onx/ssd-mobilenet.onnx.1.1.7103.GPU.FP16.engine
labelfile-path=samples/models/onx/labels.txt
force-implicit-batch-dim=1
input-dims=3;300;300;0
batch-size=1
process-mode=1
model-color-format=0
network-mode=2
num-detected-classes=2
parse-bbox-func-name=NvDsInferParseCustomSSD
custom-lib-path=/opt/nvidia/deepstream/deepstream-5.0/sources/objectDetector_SSD/nvdsinfer_custom_impl_ssd/libnvdsinfer_custom_impl_ssd.so
interval=0
gie-unique-id=1
input-blob-name = input_0
output-blob-names=boxes;scores
and the resultant output after running the deepstream app is:
Could not find NMS layer buffer while parsing
0:00:13.542989487 20826 0x1806d5e0 ERROR nvinfer gstnvinfer.cpp:613:gst_nvinfer_logger: NvDsInferContext[UID 1]: Error in NvDsInferContextImpl::fillDetectionOutput() <nvdsinfer_context_impl_output_parsing.cpp:725> [UID = 1]: Failed to parse bboxes using custom parse function
Segmentation fault (core dumped)
Is the issue with config parameters or the model convertion? what additional steps should be followed while converting the model? The convertion was done using this https://github.com/dusty-nv/pytorch-ssd/blob/e7b5af50a157c50d3bab8f55089ce57c2c812f37/onnx_export.py