We already have several yolov8 models in samples. We do not need yet another one, if it just brings fatter backbone, but not some DNN features.

But we think the model.yml file.

It says that it supports the YOLOv8L model, but there is no link for the YOLOv8L model.
Thus, it's better to support YOLOv8L to satisfy what Line 22 says.

Thanks!

Looks like the model was exported with another parameters than all other yolos we have in the sample. The sample generate a lot of garbage boxes that do not correspond to real objects. Command for the sample: ./bin/example_dnn_yolo_detector --model=./yolov8l.onnx --yolo=yolov8

Hi, the models.yml scale is 0.0039, and it's necessary to give this argument.

If not passing --scale = 0.0039:

./yolo_detecor --input=./261902.jpg --yolo=yolov8 --model=./yolov8l.onnx 

The result of yolo_detecor.cpp is

After passing --scale = 0.0039

./yolo_detecor --input=./261902.jpg --yolo=yolov8 --model=./yolov8l.onnx  --scale=0.0039 

The result becomes:

Please take a look, Thanks!

Hello, I found some mistakes when testing, so I removed the unnecessary whitespace in line92 and pushed the new commit; thanks.

Samples/dnn/models.yml:92: trailing whitespace.