Hacktoberfest 2026 : les issues que les mainteneurs ont marquées pour octobre, ouvertes et accessibles aux débutants. Parcourir les issues Hacktoberfest

Track Ultralytics stream=True inference over iterator consumption

Ouverte
  • Quelque chose ne va pas ?
#85 0 commentaires 0 réactions 0 personnes assignées Voir sur GitHub

Les mainteneurs répondent en général sous 1 jour

Personne n'a encore pris cette issue.

Évaluation

Difficulté
4/5
Temps estimé
3-5 jours
Accessibilité débutants
45/100
Type d'issue
Fonctionnalité
Clarté
Plutôt claire
Activité
Active
Stack technique
python
Domaine
observability

Piste de recherche

Read the Ultralytics adapter and the shared SyncStreamWrapper introduced in #82, then compare the eager predict() coverage in #84. Trace how streamed results are consumed and finalized; preserve lazy iteration and keep result interpretation in the adapter. Done means both entry points emit one consumption-aware event on exhaustion, close, or error, with bounded metadata and compatibility tests for supported Ultralytics versions.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

Current behavior

Ultralytics/YOLO returns a lazy results iterator for model.predict(..., stream=True) and model(..., stream=True). Inference runs as the application consumes it, but the current integration records success and duration when the iterator is returned. This can report success before inference runs, omit detection/classification output metadata, and miss failures during iteration.

PR #84 fixes eager predict() entry-point coverage; it does not address lazy streaming lifecycle tracking.

Possible approach

Reuse the shared SyncStreamWrapper introduced in #82 for iteration timing, exactly-once finalization, closure/error handling, trace correlation, and telemetry failure isolation. Keep YOLO result interpretation in the Ultralytics adapter through integration-specific callbacks.

A proposed starting contract is one inference event per streamed prediction call, finalized when the iterator is exhausted or explicitly closed, with bounded aggregation of available detection/classification metadata. Do not retain every result or consume the iterator ahead of the application. Per-frame events would be a separate design decision.

Verification when implemented

  • Creating a lazy iterator emits no premature successful inference event.
  • Exhaustion records one event with consumption-aware duration and available result metadata.
  • Early closure and iteration errors finalize once without changing application behavior.
  • Both entry points behave consistently, with no duplicate events.
  • Real supported Ultralytics versions are covered by compatibility tests.

This is a known coverage gap to track for future demand; no user has requested streaming support yet. It is not a prerequisite for the eager-call fix in #84.

Langage dominant
Python
Étoiles
17
Forks
1
Merge moyen
17 h 5 min
PR mergées (30 j)
34

Préparer son environnement

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de wild-edge/wildedge-python

Toutes les issues de wild-edge/wildedge-python

Issues similaires

Plus d'issues Python

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.