Struct ObservationModel 

pub struct ObservationModel {
    pub pixel_noise_px: f64,
    pub dropout_prob: f64,
    pub clutter_per_frame: f64,
    pub occlusion: OcclusionModel,
}

Observation-model imperfections applied to the 2D detections. All default to zero, so the perfect-world baseline is unchanged.

Everything is sampled deterministically from the scenario seed plus the (camera, frame, insect) indices, so a (config, seed) reproduces a run exactly — the harness can print the seed on failure and replay it.

Fields

pixel_noise_px: f64

Standard deviation (pixels) of zero-mean Gaussian jitter added to each projected detection. Models finite localization accuracy. Default 0.

dropout_prob: f64

Per-(camera, insect, frame) probability in [0, 1] that a real detection is missed (dropped), i.i.d. per frame. Models sporadic detector misses; for sustained misses use Self::occlusion instead. Default 0.

clutter_per_frame: f64

Expected number of spurious “clutter” detections per camera per frame (false positives, placed uniformly in the image). Stresses data association. Modeled as a fixed count floor(x) plus a fractional part included with probability x - floor(x). Default 0.

occlusion: OcclusionModel

Temporally-correlated occlusion: hides an insect from a camera for whole spans of frames (default: never). See OcclusionModel.

Implementations

impl ObservationModel

pub fn is_dropped( &self, seed: u64, cam_index: usize, fno: usize, insect_id: u32, ) -> bool

Whether a real detection of insect_id on camera cam_index at frame fno is dropped, given the scenario seed. Deterministic.

pub fn is_suppressed( &self, seed: u64, cam_index: usize, fno: usize, insect_id: u32, ) -> bool

Whether a real detection of insect_id on camera cam_index at frame fno should be suppressed for any reason — an i.i.d. Self::is_dropped miss or an OcclusionModel span. This is the single check both the image backend and the in-process injector apply before emitting a point.

pub fn jitter_pixel( &self, seed: u64, cam_index: usize, fno: usize, insect_id: u32, x: f64, y: f64, ) -> (f64, f64)

The projected pixel (x, y) with deterministic Gaussian jitter applied. Returns the input unchanged when pixel_noise_px == 0.

pub fn clutter( &self, seed: u64, cam_index: usize, fno: usize, width: usize, height: usize, ) -> Vec<(f64, f64)>

Spurious clutter detections for camera cam_index at frame fno, placed uniformly within a width x height image. Deterministic.

Trait Implementations

impl Clone for ObservationModel

fn clone(&self) -> ObservationModel

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ObservationModel

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

Formats the value using the given formatter.

impl Default for ObservationModel

fn default() -> ObservationModel

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for ObservationModel

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for ObservationModel

fn eq(&self, other: &ObservationModel) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for ObservationModel

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for ObservationModel