"""Multi-face photo recognition page.""" from __future__ import annotations from pathlib import Path from PySide6.QtCore import Qt from PySide6.QtWidgets import QSplitter, QTableWidget, QTableWidgetItem, QVBoxLayout, QWidget from ..core.exporters import export_annotated_image, export_csv, export_json from ..core.utils import read_image, save_image, timestamp_for_filename from ..widgets.table_utils import configure_table_columns, refresh_table_columns from ..widgets.upload_preview import UploadPreview from .base import BasePage class MultiFacePhotoPage(BasePage): def __init__(self, context, parent=None): super().__init__(context, "Multi-face Photo Recognition", "Detect all faces in a group photo and match them against People Library.", parent) self.image_path = "" self.image = None self.rows = [] self.photo_input = UploadPreview( "Photo", extensions=[".jpg", ".jpeg", ".png", ".bmp", ".webp"], dialog_filter="Images (*.jpg *.jpeg *.png *.bmp *.webp);;All Files (*)", ) self.photo_input.pathChanged.connect(lambda path: self.load_photo(path) if path else self.clear_photo()) self.content.addWidget( self.row( self.button("Recognize Faces", self.recognize), self.button("Export Annotated Image", self.export_image), self.button("Export CSV / JSON", self.export_rows), ) ) splitter = QSplitter(Qt.Horizontal) left = QWidget() left_layout = QVBoxLayout(left) self.viewer = self.photo_input left_layout.addWidget(self.viewer) right = QWidget() right_layout = QVBoxLayout(right) self.table = QTableWidget(0, 6) self.table.setHorizontalHeaderLabels(["face_index", "person_name", "similarity", "status", "bbox", "det_score"]) configure_table_columns(self.table, [90, 180, 90, 110, 230, 90]) right_layout.addWidget(self.table) splitter.addWidget(left) splitter.addWidget(right) self.content.addWidget(splitter, 1) def load_photo(self, path: str) -> None: self.image_path = path self.image = read_image(path) if self.image is None: self.viewer.clear(emit=False) self.show_error("Image read failure.") return self.viewer.set_image(self.image, path) def clear_photo(self) -> None: self.image_path = "" self.image = None self.rows = [] self.viewer.clear(emit=False) self.table.setRowCount(0) def recognize(self) -> None: if self.image is None: self.show_error("Please select a photo.") return if not self.context.engine.is_loaded(): self.show_error("Model is not loaded. Please open Models.") return def task(): faces = self.context.engine.detect_faces(self.image, source_path=self.image_path) gallery = self.context.storage.load_all_gallery_embeddings() rows = [] overlays = [] media_id = self.context.storage.add_media_item(self.image_path, "image", width=self.image.shape[1], height=self.image.shape[0], processed_at=timestamp_for_filename()) for index, face in enumerate(faces): result = None if face.normed_embedding is not None and gallery: result_list = self.context.storage.search_embeddings(face.normed_embedding, self.context.config.default_top_k, self.context.config.recognition_threshold) result = result_list[0] if result_list else None name = result.person_name if result and result.similarity >= self.context.config.recognition_threshold else "Unknown" similarity = result.similarity if result else 0.0 status = "matched" if result and result.similarity >= self.context.config.recognition_threshold else "unknown" crop_path = "" if face.crop is not None: crop_path = str(Path(self.context.config.crop_dir) / f"media_{media_id}_{index}_{timestamp_for_filename()}.png") save_image(crop_path, face.crop) self.context.storage.add_media_face(media_id, face.normed_embedding, crop_path=crop_path, bbox=face.bbox, kps=face.kps, det_score=face.det_score, predicted_person_id=result.person_id if result else None, similarity=similarity, status=status) row = {"face_index": index, "person_name": name, "similarity": similarity, "status": status, "bbox": face.bbox, "det_score": face.det_score, "crop_path": crop_path} rows.append(row) overlays.append({"bbox": face.bbox, "label": name, "similarity": similarity}) return rows, overlays def done(payload): self.rows, overlays = payload self.viewer.set_faces(overlays) self.table.setRowCount(len(self.rows)) columns = ["face_index", "person_name", "similarity", "status", "bbox", "det_score"] for row_index, row in enumerate(self.rows): for col, key in enumerate(columns): value = row[key] if isinstance(value, float): value = f"{value:.4f}" self.table.setItem(row_index, col, QTableWidgetItem(str(value))) refresh_table_columns(self.table) self.set_status(f"Recognized {len(self.rows)} face(s).") self.run_task("Recognizing photo", task, done) def export_image(self) -> None: image = self.viewer.render_with_overlay() if image is None: self.show_error("No annotated image to export.") return path = Path(self.context.config.export_dir) / f"annotated_{timestamp_for_filename()}.png" export_annotated_image(path, image) self.set_status(f"Annotated image exported to {path}") def export_rows(self) -> None: if not self.rows: self.show_error("No recognition results to export.") return path = Path(self.context.config.export_dir) / f"multiface_{timestamp_for_filename()}.csv" export_csv(path, self.rows) export_json(path.with_suffix(".json"), self.rows) self.set_status(f"Results exported to {path}")