qengineering / yolox-tracking-ncnn-rpi_64-bit Goto Github PK

Hi, @Qengineering

I saw that below 2 link from the top is the same link and not firecting the Tflite-tracker code.

Is these fps for tflite + tracking ?

Can raspbery pi 4 handle the vehicle count ie 20 fps ?

Hi @Qengineering ,

When I run the project file YoloX_Byte_Track.cbp, I got following error:

||=== Build: Release in YoloX Byte Track (compiler: GNU GCC Compiler) ===|

/usr/bin/ld: obj/Release/src/yoloX.o:(.data.rel.ro._ZTI11YoloV5Focus[_ZTI11YoloV5Focus]+0x10)|
|undefined reference to 'typeinfo for ncnn::Layer'|

||error: ld returned 1 exit status|

||=== Build failed: 2 error(s), 0 warning(s) (0 minute(s), 1 second(s)) ===|

I have installed Eigen and working with your RPi 64 Img. I have checked the previous #1. It seems not working for me. What is confusing me is that YoloX detection was running on the system with the code

yolox.register_custom_layer("YoloV5Focus", YoloV5Focus_layer_creator);
It means that my ncnn should be fine to work with YoloX. However, it fails in the tracking program.
I also checked the Project build options that ncnn library has been added in the compiler. I am not sure which part causes failure this time.

Hi, @Qengineering
Thanks for this great work.
I was trying to port the models to android using ndk and jni.
There is a problem with a fstream function inside bytetracker. I see "fstream" is included in bytetracker.cpp, but I had no luck finding the code using it. And just compiles OK without it until being run on android.
Is there a file operation inside this library? Do you have any suggestions?
Regards.

Hi,
Is there any way to define an interval for the detector model, And leave the tracker to do the hard work in those intervals?
similar to this issue but with a different tracker.