Function postprocess 

pub fn postprocess(
    outputs: Vec<(&[f32], Vec<usize>)>,
    task: Task,
    preprocess: &PreprocessResult,
    config: &InferenceConfig,
    names: &HashMap<usize, String>,
    orig_img: Array3<u8>,
    path: String,
    speed: Speed,
    inference_shape: (u32, u32),
    end2end: bool,
    kpt_shape: Option<(usize, usize)>,
) -> Results

Post-process raw model output based on task type.

Arguments

• outputs - Vector of raw model outputs (data, shape).
• task - The task type (detect, segment, pose, etc.).
• preprocess - Preprocessing result containing scale/padding info.
• config - Inference configuration.
• names - Class ID to name mapping.
• orig_img - Original image as HWC array.
• path - Source image path.
• speed - Timing information.
• inference_shape - The inference tensor shape (height, width) after letterboxing.

Returns

Processed Results object.