Do I need to add parse function when I deal with other yolov5 models? about deepstream-yolo HOT 6 CLOSED

Hi, I didn't understand what you need. Do you want know how to run model or how to get processed metada?

from deepstream-yolo.

Hi, I didn't understand what you need. Do you want know how to run model or how to get processed metada?

Hi,

Thank you for your prompt replay.

In your /examples/yolov5s/nvdsinfer_custom_impl_Yolo/nvdsparsebbox_Yolo.cpp file, there is a function named：NvDsInferParseCustomYoloV5(), I think it can parse yolov5s meta data. If I want to parse v5m or v5x, can I just use this function or I need create a new one specific for v5m/v5x?

from deepstream-yolo.

Do you need only TensorRT converted model to use with other yolov5 model, the nvdsparsebbox_Yolo.cpp file is the same of yolov5s.

Metadata is processed stream/video data. It's location of bbox, label of detected object, etc. To get this data, you need knowledge to edit deepstream code (Python or C).

from deepstream-yolo.

Got it, thank you for your confirmation.

from deepstream-yolo.

@theiot, I updated and simplified YOLOv5.md.

from deepstream-yolo.

@theiot, I updated and simplified YOLOv5.md.

Cool!

from deepstream-yolo.