score oplabels2d × labels2d → tableimport fullseye as fs; fs.ledger.seg_contingency(a, b, ignore_label=None) (実装を直接呼ぶなら import segcompare; segcompare.seg_contingency(a, b, ignore_label=None)、台帳から引くなら opsemproof.get("seg_contingency"))The contingency table of truth a against candidate b (sparse).
Returns labels_a / labels_b (the distinct labels), rows / cols (indices
into them) and counts for every co-occurring pair, plus n (pixels counted),
row_sums (size of each truth object) and col_sums (size of each candidate object).
counts.sum() == n, row_sums.sum() == col_sums.sum() == n.
Raises ValueError: non-integer, non-finite, boolean-free-but-string or masked input;
not 2-D / 3-D; empty; different shapes; every pixel ignored.
py -3.11 examples/gallery_mesh_seg_sun_gains.pytable を入力に取れる)seg_synapse_partners · seg_wiring_variation · seg_synapse_nri · seg_wiring_exposure
score)seg_variation_of_information · seg_rand
Provenance: segcompare.py — EMPROOF operator registry. この per-op ノートは tools/opdocs.py md が自動生成(手編集しない)。
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.