We infer the AI video stream through deepstream, and now we want to capture the image of the AI video stream, but now we can only capture the image of the original video stream, and the captured image has no information such as the boxes in the AI video stream. How to solve it, thanks.
It is not clear about the request. Please try to run deepstream-app with the config file:
/opt/nvidia/deepstream/deepstream-6.0/samples/configs/deepstream-app$ deepstream-app -c source8_1080p_dec_infer-resnet_tracker_tiled_display_fp16_nano.txt
By default video frames are rendered out. Do you mean you would like to encode the frames into h264/h265 stream and save to a file?
For example, I want to capture a picture like this：
But the screenshot method I know now is to capture such pictures：
The captured image has no information such as AI box.
Is there any way to solve it
Do you run deepstream-app? If you run deepsteam-app with the config file, you should see output like this:
Should see rectangle for detected objects.
How to implement screenshot function in deepstream
By default nvdsosd links to display sinks like nvoverlaysink or nveglglessink. You can modify the code to link like
… ! nvdsosd ! dsexample! …
and add code in dsexample to encoder and save to JPEG.
Or add prob function to source pad of nvdsosd, and add code in the prob function.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.