Preprocessing for image contrast enhancement for peoplenet

[edit_or]

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) RTX3050
• DeepStream Version 7.0

peoplenet model is used for my application.
Pls see one image of my video stream. It is night vision so image contrast is not good. I like to enhance image contrast to improve peoplenet model detection.

Looked at deepstream-preprocess-test sample and not quite sure what to do.
So to deploy preprocessing before pgie, the command is like follow? Need to add config_preprocess.txt before main config file?

./deepstream-app -c ../../../../samples/configs/config_preprocess.txt ../../../../samples/configs/deepstream-app/peoplenet_main.txt

How to implement image enhancement for contrast improvement in config file?

[Fiona.Chen]

@edit_or

So to deploy preprocessing before pgie, the command is like follow? Need to add config_preprocess.txt before main config file?

For the Gst-nvdspreprocess (Alpha) — DeepStream documentation 6.4 documentation plugin usage, please refer to the document and there is a deepstream-app sample config in /opt/nvidia/deepstream/deepstream/samples/configs/deepstream-app/source4_1080p_dec_preprocess_infer-resnet_preprocess_sgie_tiled_display_int8.txt

@edit_or

How to implement image enhancement for contrast improvement in config file?

There is no contrast enhancement algorithm in DeepStream SDK. You may need to develop the algorithm by yourself.