Could not find bbox layer buffer while parsing

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi,

In deepstream 4.0, I used deepstream-infer-tensor-meta-test sample with parse-bbox-func-name and custom-lib-path as mentioned in objectDetector_SSD sample, I used model provided in https://github.com/AastaNV/DeepStream/tree/master/Model/DetectNet

my config file,

batch-size=2
network-mode=1
process-mode=1
model-color-format=0
num-detected-classes=1
interval=0
gie-unique-id=1
output-blob-names=coverage;bboxes
parse-bbox-func-name=NvDsInferParseCustomSSD
custom-lib-path=/libnvdsinfer_custom_impl_ssd.so

I get Could not find bbox layer buffer while parsing with no bbox in video. How can I solve this?

2

Hi,

Is there any example to parse detectnet output using custom parser in deepstream 4.0?

3

Hi,

You can find the DetectNet bbox parser here:
[url]https://github.com/AastaNV/DeepStream/tree/master/parser_detectnet[/url]

Thanks.

4

when I use it with deepstream 4.0, I get segmentaion fault (core dumped) error.

5

Hi,

The default bounding box parser is for ResNet format.
Please update the parser into DetectNet:

1. Compile the parser shared in the comment #3.

2. Update the configure file to use the customized parser

[primary-gie]
...
parse-func=0
parse-bbox-func-name=parse_bbox_custom_detectnet
parse-bbox-lib-name=/home/nvidia/parser_detectnet/libnvparsebbox.so

Thanks.