s6.app.utils.rim_alg_optm

Generate ground-truth outputs for detect_outer_rim() over a StructuredDataset.

Reads frames from temp/calibration_data (by default), takes frame[‘B’][‘image’] as input to detect_outer_rim(), and writes results into another StructuredDataset at temp/dev/rim.

This script is a first step to collect reference outputs before we optimize the detect_outer_rim() implementation.

s6.app.utils.rim_alg_optm.generate(input_dir: str = 'temp/calibration_data', output_dir: str = 'temp/dev/rim', limit: int | None = None) → None

Iterate over the input dataset, run detect_outer_rim() on B-camera images, and store outputs into the output dataset.

Parameters - input_dir: path to source StructuredDataset directory (read) - output_dir: path to destination StructuredDataset directory (write) - limit: optional maximum number of frames to process

s6.app.utils.rim_alg_optm.compare(input_dir: str = 'temp/calibration_data', baseline_dir: str = 'temp/dev/rim', limit: int | None = None, tolerance_px: float = 10.0) → None

Run both detect_outer_rim (v1) and detect_outer_rim_v2 (v2) on each frame in the input dataset and compare results and runtime.

Reports detection parity, center/radius deltas, per-axis jitter across frames (abs deltas), and timing stats including speedup.

s6.app.utils.rim_alg_optm.preview(input_dir: str = 'temp/calibration_data', baseline_dir: str = 'temp/dev/rim', start: int = 0, limit: int | None = None) → None

Interactive preview: render centers and circles from v1 (baseline) and v2.

Controls:

n/space/right: next frame p/left: previous frame q/esc: quit

s6.app.utils.rim_alg_optm.main() → None