Struct CPUProcessor 

pub struct CPUProcessor { /* private fields */ }

CPUConverter implements the ImageProcessor trait using the fallback CPU implementation for image processing.

Implementations

impl CPUProcessor

pub fn materialize_segmentations( &self, detect: &[DetectBox], proto_data: &ProtoData, letterbox: Option<[f32; 4]>, ) -> Result<Vec<Segmentation>>

Materialize segmentation masks from proto data into Vec<Segmentation>.

This is the CPU-side decode step of the hybrid mask rendering path: call this to get pre-decoded masks, then pass them to draw_decoded_masks for GPU overlay. Benchmarks show this hybrid path (CPU decode + GL overlay) is faster than the fused GPU draw_proto_masks on all tested platforms.

Optimized: fused dequantization + dot product avoids a 3.1MB f32 allocation for the full proto tensor. Uses fast sigmoid approximation.

pub fn materialize_scaled_segmentations( &self, detect: &[DetectBox], proto_data: &ProtoData, letterbox: Option<[f32; 4]>, width: u32, height: u32, ) -> Result<Vec<Segmentation>>

Produce per-detection masks at (width, height) pixel resolution by upsampling the full proto plane once then cropping per bbox. Each det.bbox is assumed to be in model-input normalized coordinates (the convention used by the decoder output); when letterbox is Some, (width, height) are original-content pixel dims and the inverse letterbox transform is applied to both the bbox (for the crop region and returned Segmentation metadata) and each output pixel (for proto-plane sampling). Mask values are binary uint8 {0, 255} after thresholding sigmoid > 0.5.

Used by [ImageProcessor::materialize_masks] when the caller selects MaskResolution::Scaled.

impl CPUProcessor

pub fn new() -> Self

Creates a new CPUConverter with bilinear resizing.

pub fn new_nearest() -> Self

Creates a new CPUConverter with nearest neighbor resizing.

Trait Implementations

impl Clone for CPUProcessor

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for CPUProcessor

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for CPUProcessor

fn default() -> Self

Returns the “default value” for a type.

impl ImageProcessorTrait for CPUProcessor

fn convert( &mut self, src: &TensorDyn, dst: &mut TensorDyn, rotation: Rotation, flip: Flip, crop: Crop, ) -> Result<()>

Converts the source image to the destination image format and size. The image is cropped first, then flipped, then rotated

fn draw_decoded_masks( &mut self, dst: &mut TensorDyn, detect: &[DetectBox], segmentation: &[Segmentation], overlay: MaskOverlay<'_>, ) -> Result<()>

Draw pre-decoded detection boxes and segmentation masks onto dst.

fn draw_proto_masks( &mut self, dst: &mut TensorDyn, detect: &[DetectBox], proto_data: &ProtoData, overlay: MaskOverlay<'_>, ) -> Result<()>

Draw masks from proto data onto image (fused decode+draw).

fn set_class_colors(&mut self, colors: &[[u8; 4]]) -> Result<()>

Sets the colors used for rendering segmentation masks. Up to 20 colors can be set.

fn convert_deferred( &mut self, src: &TensorDyn, dst: &mut TensorDyn, rotation: Rotation, flip: Flip, crop: Crop, ) -> Result<()>

Like convert, but does not wait for the GPU to finish.

fn flush(&mut self) -> Result<()>

Complete all work enqueued by convert_deferred since the last flush, issuing a single GPU synchronization.

impl Send for CPUProcessor

impl Sync for CPUProcessor