Expand description
Phase 5 / V5 — multi-object detector pipeline (grid head + NMS).
Honest scope: pure-Rust, fixed-buffer path over a configurable cell grid.
Not a neural YOLO head — scores each cell by channel mean / edge energy
(same family as CpuReferenceVision) then applies class-agnostic NMS.
Replace the cell scorer with a P64 encoder when weights land; keep the
pipeline + VisualModel surface.
Structs§
- Grid
Multi Object Detector - Configurable multi-object grid detector.
Constants§
- CLASS_
HIGH_ EDGE - CLASS_
LOW_ CONTRAST - CLASS_
MOSTLY_ BLUE - CLASS_
MOSTLY_ GREEN - CLASS_
MOSTLY_ RED - Stable class IRIs (shared taxonomy with cpu_reference).
- MAX_
GRID - Maximum grid dimension (NxN cells before NMS). 8×8 = 64 = MAX_DETECTIONS.
Functions§
- sample_
frame_ indices - Deterministic frame indices for video sampling (cold path helper).
Writes every
stride-th frame intoout, starting at 0. Returns count written. Overflow: if more indices thanout.len(), only firstout.len()are kept (deterministic).