Struct RegionLearner 

pub struct RegionLearner { /* private fields */ }

Layer-learning state for one run: accumulators plus the last informative native frame.

Implementations

impl RegionLearner

pub fn new(width: usize, height: usize) -> RegionLearner

A learner over a width × height analysis grid.

pub fn accumulators(&self) -> &Learner

The accumulators, for hosts and tests that read them back.

pub fn add( &mut self, field: &Field<'_>, prev: &[u8], current: &[u8], native: Option<(&[u8], &[u8], usize, usize)>, ) -> bool

Accumulates one field (the learner applies its own informative gate). prev/current are analysis grays; native is the pair of native RGBA frames with their shared size, when the host has both. When the field was accumulated and native frames were given, current’s native frame becomes the appearance reference finish samples. Returns whether the field was accumulated.

pub fn finish( &self, native_width: f64, native_height: f64, manual: &[ManualRegion], factor: u32, ) -> Vec<RegionOut>

Builds the run’s regions. With manual regions, those (in order, each excluding every later one and every ignored one) plus an unassigned region when they leave native pixels uncovered; otherwise the automatic layout from the accumulators. factor is the integer analysis factor.