Files
aimsight-libredesk/snippets/textes/how-can-i-detect-deformation-of-the-reference-network.md
T
maximeandClaude Opus 5.5 10ed887417 Noms plus clairs : snippets/, scripts de test numérotés, verifier_ollama.py
- base_connaissances/ → snippets/ (corpus/ → textes/, instructions.md → consignes_redaction.md, decoupage.py → fichiers.py, importer.py → envoyer.py) ; python3 -m snippets.envoyer.
- instance-de-test/ : 1_demarrer.sh, 2_creer_boite_et_cle.py, 3_brancher_ia.py, 4_essayer_brouillons.py, voir_empreintes.sh, connexion_admin.py.
- ollama/verifier.py → ollama/verifier_ollama.py ; tests, tâches VS Code et docs alignés.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-12 10:27:06 +01:00

1.0 KiB

How can I detect deformation of the reference network?

Camera-motion estimation provides a reprojection error called sigma0, which is available in the system metrics. Sigma0 measures how well the observed displacement of the different reference targets can be explained by the estimated global camera motion. An abnormal increase in sigma0 can therefore indicate that the reference network is deforming in a way that can no longer be explained by simple camera movement. A low value indicates that the observed movements of the reference targets remain consistent with a common camera movement. It is useful for monitoring the consistency and stability of the reference network and can be used to trigger further investigation if necessary. Sigma0 and the per-target Error Value provide complementary information: • Error Value checks whether an individual target is still visible and resembles its reference image; • sigma0 checks whether the reference targets remain geometrically consistent with the estimated camera movement.

Source : AIMsight FAQ, rév. 1.5