Function decode_semantic 

pub fn decode_semantic(
    input: &TensorDyn,
    config: &SemanticConfig,
    dst: &mut TensorDyn,
    confidence: Option<&mut TensorDyn>,
    options: &SemanticOptions,
) -> DecoderResult<()>

Decodes a semantic output into dst, an [H, W] class map of dtype u8 (up to 256 classes) or u16 (up to 65 536).

confidence, when given, receives the winning class’s probability as an [H, W] map of f32 in [0, 1] or u8 scaled to 255. It is computed at the score map’s resolution and upsampled, so it is approximate and meant for overlays. Class-map inputs have no confidence.

Errors

• DecoderError::InvalidShape if input does not have the configured shape, or dst and confidence are not matching [H, W] tensors.
• DecoderError::NotSupported for unsupported dtypes, or when the class count exceeds what the dtype of dst holds.
• DecoderError::InvalidConfig for an invalid letterbox or quantization, or a class-map value dst cannot hold.
• DecoderError::Tensor carrying CpuAccessDenied when input is not declared CPU-readable, or dst or confidence not CPU-writable.