enh: Split off face embeddings into a deduplicated table

lib/Classifiers/Images/ClusteringFaceClassifier.php

@@ -9,8 +9,9 @@
 
 use OCA\Recognize\BackgroundJobs\ClusterFacesJob;
 use OCA\Recognize\Classifiers\Classifier;
-use OCA\Recognize\Db\FaceDetection;
 use OCA\Recognize\Db\FaceDetectionMapper;
+use OCA\Recognize\Db\FaceEmbedding;
+use OCA\Recognize\Db\FaceEmbeddingMapper;
 use OCA\Recognize\Service\Logger;
 use OCA\Recognize\Service\QueueService;
 use OCP\AppFramework\Services\IAppConfig;
@@ -41,6 +42,7 @@ public function __construct(
 		Logger $logger,
 		IAppConfig $config,
 		private FaceDetectionMapper $faceDetections,
+		private FaceEmbeddingMapper $faceEmbeddings,
 		QueueService $queue, IRootFolder $rootFolder,
 		private IUserMountCache $userMountCache, private IJobList $jobList, ITempManager $tempManager, IPreview $previewProvider,
 		private IManager $shareManager) {
@@ -73,7 +75,7 @@ public function classify(array $queueFiles): void {
 		$filteredQueueFiles = [];
 		foreach ($queueFiles as $queueFile) {
 			try {
-				$facesByFileCount = count($this->faceDetections->findByFileId($queueFile->getFileId()));
+				$facesByFileCount = $this->faceEmbeddings->countByFileId($queueFile->getFileId());
 			} catch (Exception $e) {
 				$this->logger->debug('finding faces by file '.$queueFile->getFileId().' failed', ['exception' => $e]);
 				$facesByFileCount = 1;
@@ -114,19 +116,25 @@ public function classify(array $queueFiles): void {
 					$userIds = [];
 				}
 
+				$embedding = new FaceEmbedding();
+				$embedding->setX($face['x']);
+				$embedding->setY($face['y']);
+				$embedding->setWidth($face['width']);
+				$embedding->setHeight($face['height']);
+				$embedding->setVector($face['vector']);
+				$embedding->setFileId($queueFile->getFileId());
+				try {
+					$this->faceEmbeddings->insert($embedding);
+				} catch (\Throwable $e) {
+					$this->logger->error('Could not store face embedding in database', ['exception' => $e]);
+					continue;
+				}
+
 				// Insert face detection for all users with access
 				foreach ($userIds as $userId) {
 					$this->logger->debug('preparing face detection for user ' . $userId);
-					$faceDetection = new FaceDetection();
-					$faceDetection->setX($face['x']);
-					$faceDetection->setY($face['y']);
-					$faceDetection->setWidth($face['width']);
-					$faceDetection->setHeight($face['height']);
-					$faceDetection->setVector($face['vector']);
-					$faceDetection->setFileId($queueFile->getFileId());
-					$faceDetection->setUserId($userId);
 					try {
-						$this->faceDetections->insert($faceDetection);
+						$this->faceDetections->insertForUser($embedding, $userId);
 					} catch (\Throwable $e) {
 						$this->logger->error('Could not store face detection in database', ['exception' => $e]);
 						continue;

lib/Dav/Faces/FacePhoto.php

@@ -46,11 +46,15 @@ public function getName(): string {
 
 	#[Override]
 	public function delete(): void {
-		$detections = $this->detectionMapper->findByClusterId($this->faceDetection->getClusterId());
-		if (count($detections) > 1) {
+		$detections = $this->detectionMapper->findByClusterIdWithVector($this->faceDetection->getClusterId());
+		// $this->faceDetection was loaded without its vector, so take it from the cluster's detections
+		$thisDetection = current(array_filter($detections, fn (FaceDetection $detection) => $detection->getId() === $this->faceDetection->getId()));
+		if (count($detections) > 1 && $thisDetection !== false) {
 			$centroid = FaceClusterAnalyzer::calculateCentroidOfDetections($detections);
 			$distance = new Euclidean();
-			$distanceValue = $distance->compute($centroid, $this->faceDetection->getVector());
+			/** @var list<float> $vector */
+			$vector = $thisDetection->getVector();
+			$distanceValue = $distance->compute($centroid, $vector);
 			// Set threshold to avoid recreating the same mistake
 			$this->faceDetection->setThreshold($distanceValue);
 		}

lib/Db/FaceDetection.php

@@ -29,6 +29,11 @@
  * @method int|null getClusterId()
  * @method float getThreshold()
  * @method setThreshold(float $threshold)
+ * @method int|null getEmbeddingId()
+ * @method setEmbeddingId(int $embeddingId)
+ *
+ * The face vector is stored once per file in {@see FaceEmbedding} and only populated
+ * on detections loaded with {@see FaceDetectionMapper} methods that join it in.
  */
 class FaceDetection extends Entity {
 	protected $fileId;
@@ -37,17 +42,18 @@ class FaceDetection extends Entity {
 	protected $y;
 	protected $height;
 	protected $width;
+	protected $embeddingId;
 	protected $faceVector;
 	protected $clusterId;
 	protected $threshold;
 	/**
 	 * @var string[]
 	 */
-	public static $columns = ['id', 'user_id', 'file_id', 'x', 'y', 'height', 'width', 'face_vector', 'cluster_id', 'threshold'];
+	public static $columns = ['id', 'user_id', 'file_id', 'x', 'y', 'height', 'width', 'embedding_id', 'cluster_id', 'threshold'];
 	/**
 	 * @var string[]
 	 */
-	public static $fields = ['id', 'userId', 'fileId', 'x', 'y', 'height', 'width', 'faceVector', 'clusterId', 'threshold'];
+	public static $fields = ['id', 'userId', 'fileId', 'x', 'y', 'height', 'width', 'embeddingId', 'clusterId', 'threshold'];
 
 	public function __construct() {
 		// add types in constructor
@@ -57,6 +63,7 @@ public function __construct() {
 		$this->addType('y', 'float');
 		$this->addType('height', 'float');
 		$this->addType('width', 'float');
+		$this->addType('embeddingId', 'integer');
 		$this->addType('faceVector', 'json');
 		$this->addType('clusterId', 'integer');
 		$this->addType('threshold', 'float');
@@ -65,7 +72,7 @@ public function __construct() {
 	public function toArray(): array {
 		$array = [];
 		foreach (static::$fields as $field) {
-			if ($field === 'faceVector') {
+			if ($field === 'embeddingId') {
 				continue;
 			}
 			$array[$field] = $this->{$field};
@@ -79,4 +86,13 @@ public function getVector(): array {
 	public function setVector(array $vector): void {
 		$this->setter('faceVector', [$vector]);
 	}
+
+	/**
+	 * The vector is not a column of the detections table
+	 */
+	public function getUpdatedFields(): array {
+		$fields = parent::getUpdatedFields();
+		unset($fields['faceVector']);
+		return $fields;
+	}
 }