Function impl_yolo_float 

pub fn impl_yolo_float<B: BBoxTypeTrait, T: Float + AsPrimitive<f32> + Send + Sync>(
    output: ArrayView2<'_, T>,
    score_threshold: f32,
    iou_threshold: f32,
    nms: Option<Nms>,
    output_boxes: &mut Vec<DetectBox>,
)
where
    f32: AsPrimitive<T>,

Internal implementation of YOLO decoding for float tensors.

Expected shapes of inputs:

• output: (4 + num_classes, num_boxes)