Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU) Jetson
• DeepStream Version 5.0
• JetPack Version (valid for Jetson only) 4.4
• TensorRT Version 7
• NVIDIA GPU Driver Version (valid for GPU only)
helo everyone!
my pipeline is base on test2, like this:
source->h264parser->decoder->streammux->pgie->nvtracker(nvdcf)->nvnidconv->caps_filter->dsexample->sink,
pipeline is slow when i tracking too many object(over 40).
i want to filter object by pgie, then send it to nvtracker, so i can Improve fps.
but I don’t know how to do it. please help me.
What do you mean “too many object(over 40)”? Are these 40 objects you want to detect? What kind of objects do you want to filter out?
There are several ways in nvinfer to filter out some kind of objects:
- Filter with size.
With “detected-min-w”, “detected-min-h”, “detected-max-w” and “detected-max-h” configurations in nvinfer config file, the detected objects size larger than the max or less than the min will be filtered out.
- Filter with ROI
The “roi-top-offset” and “roi-bottom-offset” configuration in nvinfer config file will filter according to the area. If the object appear outside ROI area, it will be filtered out.
- Filter with class id
The “filter-out-class-ids” configuration in nvinfer config file will filter out specific classes of objects. All objects listed in “filter-out-class-ids” will be filtered out.
https://docs.nvidia.com/metropolis/deepstream/dev-guide/index.html#page/DeepStream%20Plugins%20Development%20Guide/deepstream_plugin_details.html#wwpID0E04DB0HA