deepinsight/insightface · error · ValueError

1:N evaluation requires at least one known probe image.

Error message

1:N evaluation requires at least one known probe image.

What it means

Raised by run_identity_identification_evaluation when no known probe images were collected (probe_items is empty). Probes with known identity are required to compute top-1/top-N accuracy; without them the evaluation has no ground truth to score.

Source

Thrown at python-package/insightface/gui/core/evaluation.py:840

def run_identity_identification_evaluation(
    dataset_root: str | Path,
    engine: FaceEngine,
    auto_split: bool = False,
    multi_face_policy: str = MULTI_FACE_REQUIRE_ONE,
    license_status: str = DEFAULT_LICENSE_STATUS,
    progress_callback=None,
    cancel_callback=None,
) -> EvaluationResult:
    root = Path(dataset_root).expanduser()
    if not root.is_dir():
        raise ValueError(f"Dataset root not found: {root}")
    gallery_items, probe_items, unknown_items = (
        _collect_gallery_probe_auto_split(root) if auto_split else _collect_gallery_probe_structured(root)
    )
    if not gallery_items:
        raise ValueError("1:N evaluation requires at least one gallery image.")
    if not probe_items:
        raise ValueError("1:N evaluation requires at least one known probe image.")

    errors: List[Dict[str, Any]] = []
    cache: Dict[str, np.ndarray] = {}
    gallery: List[Dict[str, Any]] = []
    start_all = time.perf_counter()
    for index, item in enumerate(gallery_items):
        embedding = _embedding_for_image(
            Path(item["path"]),
            engine,
            cache,
            errors,
            "gallery",
            multi_face_policy=multi_face_policy,
        )
        if embedding is not None:
            gallery.append({"identity": item["identity"], "path": str(item["path"]), "embedding": embedding})
        if progress_callback:
            progress_callback(index + 1, len(gallery_items), f"Indexed gallery {index + 1}/{len(gallery_items)}")

View on GitHub (pinned to 7fadd420c2)

Solutions

1. Ensure probe/<identity>/ folders contain at least one valid image each
2. For auto_split, provide enough images per identity so the probe split is non-empty
3. Verify image extensions match what list_images accepts
4. Inspect collected counts before running by debugging _collect_gallery_probe_* output

from pathlib import Path
EXTS = {".jpg", ".jpeg", ".png", ".bmp"}
probe_dir = Path(root) / "probe"
n = sum(1 for f in probe_dir.rglob("*") if f.suffix.lower() in EXTS)
assert n > 0, "no known probes"

try:
    result = run_identity_identification_evaluation(root, auto_split=False)
except ValueError as e:
    if "at least one known probe" in str(e):
        add_probe_images()

When it happens

Trigger: Structured mode: probe/ folder missing (caught earlier if gallery also missing) or empty of images. Auto-split: too few images per identity so nothing is allocated to the probe split.

Common situations: probe folder present but images filtered out (wrong extensions, corrupt files); auto_split ratio leaving all images in gallery; misnamed probe directory; test fixture forgetting probe images.

Related errors

• 1:N evaluation requires at least one gallery image.
• No verification pairs could be generated from the selected i
• 1:N without Auto Split requires gallery/ and probe/ folders.
• Dataset root not found: {root}
• Identity folder root not found: {root}

AI-assisted analysis of deepinsight/insightface@7fadd420c2 (2026-08-28). Data as JSON: /api/errors/a5a2fe544f4d18af. Report an issue: GitHub.