cascade yolo models

Accelerated Computing Intelligent Video Analytics DeepStream SDK
#1

hi,

how is it possible to adapt objectDetector_yolo example to the deepstream-test2 example in the sample apps.

i have 3 different yolo models, where i need to pass the output of one as input to the other. i have explored test-app 2 and yolo object detector in the deepstream package, but it is not clear for me how to achieve cascaded inference with yolo.

#2

You can refer to config file in https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps/tree/master/back-to-back-detectors
The nvinfer property introduction: https://docs.nvidia.com/metropolis/deepstream/plugin-manual/index.html#page/DeepStream_Plugin_Manual%2Fdeepstream_plugin_details.02.01.html

#3

test app 2 ran with yolo, but there is no bounding boxes or anything showing on the video output. how can i capture inference results and render it on the video output?

#4

hi,

i have successfully ran a custom yolo app with test app-2, but i have two issues:

1- now i need to capture inference info for each frame such as objects and their bounding box coordinates. how can i achieve this.

2- how can i make it run with an rtsp stream instead of a video file.

thanks

#5

no update on this?

#6

2- how can i make it run with an rtsp stream instead of a video file.
You can refer to test3, change uri to rtsp uri

#7

hi chris,

can you elaborate? i have changed source = gst_element_factory_make (“filesrc”, “file-source”); to–>
source = gst_element_factory_make (“uridecodebin”, “uri-decode-bin”);

and g_object_set (G_OBJECT (source), “location”, argv[1], NULL); to–> g_object_set (G_OBJECT (source), “rtsp uri”, argv[1], NULL);

and got the following error:

(deepstream-test1-app:10545): GLib-GObject-WARNING **: 13:29:06.752: g_object_set_is_valid_property: object class 'GstURIDecodeBin' has no property named 'rtsp uri'
Elements could not be linked: 1. Exiting.

i also tried changing the source before the argument from ‘location’ to ‘uri’

and i got :

Elements could not be linked: 1. Exiting.
#8

Can you run test3 ? Please read test3 code and debug. Only changing “filesrc” to “uridecodebin” is not enough.

#9

hi：
Did your first question be solved, did you get the reasoning information to capture each frame? I am also using the test2 cascade yolo inference, and I cannot get the result of the inference.

#10

hi,

No, not yet. i will get back to work on it soon. what i seem to know is that i need to customize it through the dsexample plugin.

if you figure anything out please let me know.

#11

Hi ibondokji,

We haven’t heard back from you in a couple weeks, so marking this topic closed.
Please open a new forum issue when you are ready and we’ll pick it up there.

#12