YOLOv8

This version of YOLOv8 has been converted to run on the Axera NPU using w8a16 quantization.

This model has been optimized with the following LoRA:

Compatible with Pulsar2 version: 3.4

Convert tools links:

For those who are interested in model conversion, you can try to export axmodel through

The repo of AXera Platform, which you can get the detial of guide
Pulsar2 Link, How to Convert ONNX to axmodel

Support Platform

AX650
- M4N-Dock(爱芯派Pro)
- M.2 Accelerator card
AX630C

Chips	yolov8s
AX650	3.6 ms
AX630C	TBD ms

How to use

Download all files from this repository to the device


root@ax650:~/YOLO11-Pose# tree
.
|-- ax650
|   `-- yolov8s.axmodel
|-- ax_yolov8
|-- football.jpg
`-- yolov8_out.jpg

Inference

Input image:

Inference with AX650 Host, such as M4N-Dock(爱芯派Pro)

root@ax650:~/samples/AXERA-TECH/YOLOv8# ./ax_yolov8 -m ax650/yolov8s.axmodel -i football.jpg
--------------------------------------
model file : ax650/yolov8s.axmodel
image file : football.jpg
img_h, img_w : 640 640
--------------------------------------
Engine creating handle is done.
Engine creating context is done.
Engine get io info is done.
Engine alloc io is done.
Engine push input is done.
--------------------------------------
post process cost time:3.96 ms
--------------------------------------
Repeat 1 times, avg time 3.64 ms, max_time 3.64 ms, min_time 3.64 ms
--------------------------------------
detection num: 7
 0:  93%, [ 757,  215, 1131, 1156], person
 0:  93%, [   0,  354,  311, 1104], person
 0:  93%, [1351,  342, 1627, 1032], person
 0:  91%, [ 488,  478,  661,  998], person
32:  87%, [ 773,  889,  829,  939], sports ball
32:  77%, [1231,  876, 1280,  922], sports ball
 0:  60%, [1840,  690, 1906,  809], person
--------------------------------------

Output image:

Inference with M.2 Accelerator card

(base) axera@raspberrypi:~/lhj/YOLOv8 $ ./axcl_aarch64/axcl_yolov8 -m ax650/yolov8s.axmodel -i football.jpg 
--------------------------------------
model file : ax650/yolov8s.axmodel
image file : football.jpg
img_h, img_w : 640 640
--------------------------------------
axclrtEngineCreateContextt is done. 
axclrtEngineGetIOInfo is done. 

grpid: 0

input size: 1
    name:   images 
        1 x 640 x 640 x 3


output size: 3
    name: /model.22/Concat_output_0 
        1 x 80 x 80 x 144

    name: /model.22/Concat_1_output_0 
        1 x 40 x 40 x 144

    name: /model.22/Concat_2_output_0 
        1 x 20 x 20 x 144

==================================================

Engine push input is done. 
--------------------------------------
post process cost time:0.98 ms 
--------------------------------------
Repeat 1 times, avg time 3.75 ms, max_time 3.75 ms, min_time 3.75 ms
--------------------------------------
detection num: 7
 0:  93%, [ 757,  215, 1131, 1156], person
 0:  93%, [   0,  354,  311, 1104], person
 0:  93%, [1351,  342, 1627, 1032], person
 0:  91%, [ 488,  478,  661,  998], person
32:  87%, [ 773,  889,  829,  939], sports ball
32:  77%, [1231,  876, 1280,  922], sports ball
 0:  60%, [1840,  690, 1906,  809], person
--------------------------------------

Output image:

AXERA-TECH
/

YOLOv8