search

marcoslucianops / DeepStream-Yolo

NVIDIA DeepStream SDK 7.0 / 6.4 / 6.3 / 6.2 / 6.1.1 / 6.1 / 6.0.1 / 6.0 / 5.1 implementation for YOLO models
MIT License
1.39k stars 344 forks source link

difference python biding and deepstream_app_config.txt #444

Open khosrooo opened 10 months ago

khosrooo commented 10 months ago

Hi Sir, i have a little confuse , i get kittie-out put in python with " f.write("[{}]".format(",".join(str(x) for x in [obj_meta.confidence, int(rect_params.left), int(rect_params.top), int(rect_params.width+rect_params.left), int(rect_params.height+rect_params.top)])))" and save in txt files but when compare with txt files created by deepstream_app_config.txt (not python) , i see the python biding can not detect in frame 0,1,2,3,4 and txt file is empthy and detection is weaker than deepstream_app_config.txt (for example number of detection is lower) do u have idea about that , it is related the python biding dose not use mp4 and used h264 format ??, Thank u for supported.

marcoslucianops commented 10 months ago

Are you using exactly the same config in both cases? I mean the same config_infer file, same streammux options, with/without tracker in both, etc? Probably, you aren't getting the first 5 frames due to the tracker age (it takes few frames to set the object as true object in the meta and the number of frames it takes depends on its configuration in the yaml file).

khosrooo commented 10 months ago

Thanku for responsing Sir. I used same config_infer_primary.txt in both of them . Could u say me please which parameter must be change in yaml file tracker until the 5 frame dose not lost

marcoslucianops commented 10 months ago

It's the probationAge parameter. Set it to 0.