README

Face Detection & Embedding Guidelines

Last Updated: March 3, 2026

Overview

This document captures the current state of PhotoPrism's face detection and embedding pipeline following the October 2025 optimizations. It should be used as the canonical reference when assessing detection quality, tuning configuration, or integrating downstream tooling that depends on FaceNet embeddings.

Key changes:

• Multi-angle scanning is enabled by default and can be tuned via configuration.
• Detection thresholds were relaxed to improve recall, while overlap handling was adjusted to preserve historical behaviour.
• All face embeddings are now L2-normalized at creation, midpoint calculation, and deserialization time to keep cosine and Euclidean comparisons consistent.
• Benchmarks were added to track the cost of hotspot routines (Embedding.Dist and EmbeddingsMidpoint).

TODO: Persist detector provenance in FaceSrc (e.g., use entity.SrcONNX for SCRFD detections) so hybrid libraries can toggle background filtering per embedding source when upgrading from Pigo.

Detection Pipeline

PhotoPrism now supports two interchangeable detection engines:

• Pigo — CPU-only cascade classifier, retains historical behaviour.
• ONNX SCRFD 0.5g — ONNX Runtime-backed CNN that delivers higher recall on occluded or off-axis faces. The ONNX engine consumes 720 px thumbnails (model input 640 px), schedules work on the meta/vision workers, and defaults to half the available CPUs (minimum 1 thread). The engine is enabled automatically when FACE_ENGINE=auto and the bundled SCRFD model is present (the prebuilt runtime targets glibc ≥ 2.27 on x86_64/arm64). Operators can switch at runtime via photoprism --face-engine=<auto|pigo|onnx> or photoprism faces reset --engine=<auto|pigo|onnx> for a full re-index.

Runtime selection lives in Config.FaceEngine(); auto resolves to ONNX when the SCRFD assets are available, otherwise Pigo. Scheduling is controlled by the face model entry in vision.yml: Config.FaceEngineRunType() simply forwards to vision.Config.RunType(ModelTypeFace) and returns never if no detector is configured. This keeps face detection aligned with embedding generation so both always run together.

Angle Sweep

• The detector now evaluates the Pigo cascade at -0.3, 0, and +0.3 radians. These angles are exposed via the new FACE_ANGLE option.
• Configuration entry points:
  • CLI flag: --face-angle=<rad> (repeatable).
  • Environment variable: FACE_ANGLE (comma-separated list).
  • Options API: Config.FaceAngles().
• At start-up the detector receives a copy of face.DetectionAngles, so runtime overrides do not mutate the global defaults.

Quality & Overlap Thresholds

• The dynamic quality curve in face.QualityThreshold was flattened for better small-face recall:
  • +12 for scales <26, +8 for <32, +6 for <40, +4 for <50, +2 for <80, +1 for <110.
• The face overlap floor remains 42 % to preserve legacy marker behaviour (OverlapThresholdFloor = 41). Tests rely on that value (e.g., Markers.Contains/SameFace).

Landmark Handling

• Landmarks are only evaluated when both eyes are successfully detected for a given face. Eye candidates and cascades respect the configurable perturbation budget.
• The primary detection angles (FACE_ANGLE) do not affect landmark estimation, which continues to run at 0° to match the cascade assumptions.

Embedding Handling

Memory Management

FaceNet embeddings are generated through TensorFlow bindings that allocate tensors in C memory. Those allocations are released by Go GC finalizers, so long-running indexing jobs can show steadily rising RSS even when the Go heap stays small. To keep memory bounded during extended face indexing runs, PhotoPrism now triggers periodic garbage collection and returns freed C-allocated tensor buffers to the OS. You can tune this behavior with PHOTOPRISM_TF_GC_EVERY (default 200; set to 0 to disable). Lower values reduce peak RSS but increase GC overhead and can slow indexing, so keep the default unless memory pressure is severe.

Normalization

All embeddings, regardless of origin, are normalized to unit length (‖x‖₂ = 1):

• NewEmbedding normalizes the raw float32 inference output.
• EmbeddingsMidpoint normalizes each contributor, averages component-wise, and renormalizes the centroid.
• UnmarshalEmbedding and UnmarshalEmbeddings normalize data when loading from persisted JSON.
• Static datasets (children/background samples) and random generators now normalize their entries after perturbation.
• photoprism faces audit --fix re-normalizes persisted embeddings, rekeys face IDs, and re-links markers (ID + FaceDist) so historical data adopts the canonical unit-length vectors.
• Faces.Match pre-filters matchable clusters, keeps an in-memory veto list for freshly cleared markers, and caches embeddings to avoid redundant distance checks; BenchmarkSelectBestFace (1024 faces) now reports a bucket size of ~16 candidates out of 1024 (≈98 % fewer distance evaluations) at ≈0.55 ms/op with zero allocations.
• Face clusters update their sample statistics (Samples, ClusterRadius) from the latest matches via Face.UpdateMatchStats, avoiding stale radii during optimize loops. The radius is capped at 0.42 so automatic matches accept new embeddings up to ClusterRadius + MatchDist (≈0.88) away from the centroid.
• Child and background embeddings remain opt-in for automated matching. Set PHOTOPRISM_FACE_MATCH_CHILDREN=true (or the FaceMatchChildren option) to include children, and PHOTOPRISM_FACE_MATCH_BACKGROUND=true to include background samples; both default to false so operators explicitly choose when these categories participate.
• Cluster materialisation now pre-sizes buffers; BenchmarkClusterMaterialize reports ~14.8 µs/op with 64 allocations (≈56 KB) versus the legacy ~29.8 µs/op with 384 allocations (≈105 KB).

This guarantees that Euclidean distance comparisons are equivalent to cosine comparisons, aligning our thresholds with FaceNet literature.

Face Kind Reference

Manual Cluster Merging & Retained Markers

The Faces.Optimize loop still prefers the operator-curated clusters (face_src = 'manual'). When multiple manual clusters for the same subject can be merged, query.MergeFaces materialises a midpoint cluster and reassigns markers to it. If some markers remain attached to the original clusters (for example because their embeddings sit far from the midpoint), the old clusters cannot be purged and the optimiser now emits a warning:

faces: retained manual clusters after merge: kept 4 candidate cluster(s) [...] for subject <uid> because markers still reference them

This is informational—the optimiser skips that merge and progresses. To reduce noise, consider:

• Running photoprism faces reset --engine=<pigo|onnx> to regenerate markers with consistent embeddings.
• Reviewing the subject’s manual clusters in the UI and trimming outliers or reassigning photos to other people.
• Confirming that the remaining clusters genuinely represent different appearances (lighting, age); in that case it is safe to ignore the warning.

No automatic data cleanup runs in this scenario, so operators remain in control of manual edits.

Additional safeguards were introduced in October 2025 so stubborn clusters are only retried a limited number of times:

• Every manual cluster now stores a retry counter (faces.merge_retry) and optional note (merge_notes). The optimiser skips clusters once the retry count reaches MergeMaxRetry (default 1). The limit may be raised or disabled with the environment variable PHOTOPRISM_FACE_MERGE_MAX_RETRY (0 = unlimited retries).
• Warnings surface only when the retry counter is incremented. Subsequent optimise runs log at debug level until counters are reset.
• photoprism faces optimize --retry clears retry counters before running the optimiser, allowing administrators to reprocess clusters after manual cleanup.
• photoprism faces audit --subject=<uid> focuses the audit report on a specific person and prints retry counts, sample statistics, and outstanding clusters so operators know which photos still need attention.
• The warning text now includes the retry count and cluster IDs.

Midpoint Computation

• The midpoint routine now performs a single pass (with vector normalization) and uses an inlined L2 distance when computing the sample radius.
• Benchmarked at ~99 µs/op and 4 KB/op for 128 vectors @512 dims, down from ~194 µs/op and >500 KB/op.

Distance Function

• Embedding.Dist was hand-optimized with loop unrolling (4-way accumulation) and now runs at ~155 ns/op, down from ~242 ns/op (≈36 % faster).
• Euclidean distance remains the recommended metric; with unit vectors, cosine similarity would yield identical rankings, so no change is required to distance thresholds.

FaceNet Integration Recommendations

• Ensure FaceNet inference remains disabled only when explicitly configured (PHOTOPRISM_FACENET_DISABLED).
• If you expose similarity scores, convert Euclidean distance to cosine using: cos θ = 1 - (d² / 2) (since embeddings are normalized).
• Keep distance thresholds (e.g., merge, clustering) expressed in the Euclidean domain unless downstream tooling mandates cosine values. The current merge tests expect distances around 0.040 for identical subjects.
• When updating pretrained models or embedding datasets, re-run the dedicated benchmarks and fixture-based tests:
  • BenchmarkEmbeddingDist
  • BenchmarkEmbeddingsMidpoint
  • TestMergeFaces/SameSubjects
  • TestNet

Troubleshooting FaceNet Model Files

If FaceNet unit tests fail with Read less bytes than requested, the local model file is typically incomplete or corrupted (assets/models/facenet/saved_model.pb).

Recovery steps:

• rm -f /tmp/photoprism/facenet.zip
• rm -rf assets/models/facenet
• make dep-tensorflow (or scripts/download-facenet.sh)
• Re-run go test ./internal/ai/face -run TestNet -count=1

Configuration Summary

Run scheduling is configured through the face model entry in vision.yml. Adjust the model’s Run value (for example on-schedule, manual, or never) to control when detection and embedding jobs execute—no separate FACE_ENGINE_RUN flag is required. When the model is left on the default auto run mode, face detection participates in manual, auto, and on-demand workflows but skips scheduled cron runs so background jobs do not trigger unexpectedly; the same applies to an explicit on-demand run mode, which now skips cron executions by default. Set Run to on-schedule explicitly if you want faces processed during scheduled vision passes.

Additional merge tuning: set PHOTOPRISM_FACE_MERGE_MAX_RETRY to control how often manual clusters are retried (default 1, 0 = unlimited). See the optimiser notes above.

Benchmark Reference

Re-run these benchmarks after any detector or embedding adjustments to catch regressions early.