Darknet: Just one CPU core is being used for object detection in darknet tiny yolo on Raspberry pi 3 B+

[Vikalp-Reorder]

I tried running tiny YOLO on raspberry pi 3 B+ for object detection from an image and it took a pretty long time to process it like 32 secs for a single image. Checked the CPU processes and found out only one CPU core was being used out of 4, which was running on a full 100% which might be slowing down the process. Is there any way that all the 4 CPU cores can be used together to speed up the detection process in Darknet Tiny Yolo?

Raspberry pi 3 B+
Raspibian

[gplast]

Hello,

You can have a look at this repository, where they optimised darknet code for Raspberry Pi Architecture.
https://github.com/digitalbrain79/darknet-nnpack

Hope that's what your looking for.

[Vikalp-Reorder]

Thanks a lot @gplast . Was looking exactly for this. Works like a charm.

[goms12]

@Vikalp-Reorder just to make sure you only use raspberry without external device like movidiusNCS etc?