Adding a file and feature to delete all detections from database

src/database/db_manager.py

@@ -462,6 +462,22 @@ class DatabaseManager:
         finally:
             conn.close()
 
+    def delete_all_detections(self) -> int:
+        """Delete all detections from the database.
+
+        Returns:
+            Number of rows deleted.
+        """
+
+        conn = self.get_connection()
+        try:
+            cursor = conn.cursor()
+            cursor.execute("DELETE FROM detections")
+            conn.commit()
+            return cursor.rowcount
+        finally:
+            conn.close()
+
     # ==================== Statistics Operations ====================
 
     def get_detection_statistics(

src/database/schema.sql

@@ -55,10 +55,7 @@ CREATE TABLE IF NOT EXISTS object_classes (
 
 -- Insert default object classes
 INSERT OR IGNORE INTO object_classes (class_name, color, description) VALUES
-    ('cell', '#FF0000', 'Cell object'),
+    ('terminal', '#FFFF00', 'Axion terminal');
-    ('nucleus', '#00FF00', 'Cell nucleus'),
-    ('mitochondria', '#0000FF', 'Mitochondria'),
-    ('vesicle', '#FFFF00', 'Vesicle');
 
 -- Annotations table: stores manual annotations
 CREATE TABLE IF NOT EXISTS annotations (

src/gui/tabs/results_tab.py

@@ -66,6 +66,13 @@ class ResultsTab(QWidget):
         self.refresh_btn.clicked.connect(self.refresh)
         controls_layout.addWidget(self.refresh_btn)
 
+        self.delete_all_btn = QPushButton("Delete All Detections")
+        self.delete_all_btn.setToolTip(
+            "Permanently delete ALL detections from the database.\n" "This cannot be undone."
+        )
+        self.delete_all_btn.clicked.connect(self._delete_all_detections)
+        controls_layout.addWidget(self.delete_all_btn)
+
         self.export_labels_btn = QPushButton("Export Labels")
         self.export_labels_btn.setToolTip(
             "Export YOLO .txt labels for the selected image/model run.\n"
@@ -139,6 +146,41 @@ class ResultsTab(QWidget):
         layout.addWidget(splitter)
         self.setLayout(layout)
 
+    def _delete_all_detections(self):
+        """Delete all detections from the database after user confirmation."""
+        confirm = QMessageBox.warning(
+            self,
+            "Delete All Detections",
+            "This will permanently delete ALL detections from the database.\n\n"
+            "This action cannot be undone.\n\n"
+            "Do you want to continue?",
+            QMessageBox.Yes | QMessageBox.No,
+            QMessageBox.No,
+        )
+
+        if confirm != QMessageBox.Yes:
+            return
+
+        try:
+            deleted = self.db_manager.delete_all_detections()
+        except Exception as exc:
+            logger.error(f"Failed to delete all detections: {exc}")
+            QMessageBox.critical(
+                self,
+                "Error",
+                f"Failed to delete detections:\n{exc}",
+            )
+            return
+
+        QMessageBox.information(
+            self,
+            "Delete All Detections",
+            f"Deleted {deleted} detection(s) from the database.",
+        )
+
+        # Reset UI state.
+        self.refresh()
+
     def refresh(self):
         """Refresh the detection list and preview."""
         self._load_detection_summary()

src/utils/create_mask_from_detection.py

@@ -0,0 +1,103 @@
+import numpy as np
+
+from pathlib import Path
+from skimage.draw import polygon
+from tifffile import TiffFile
+
+from src.database.db_manager import DatabaseManager
+
+
+def read_image(image_path: Path) -> np.ndarray:
+    metadata = {}
+    with TiffFile(image_path) as tif:
+        image = tif.asarray()
+        metadata = tif.imagej_metadata
+    return image, metadata
+
+
+def main():
+
+    polygon_vertices = np.array([[10, 10], [50, 10], [50, 50], [10, 50]])
+    image = np.zeros((100, 100), dtype=np.uint8)
+    rr, cc = polygon(polygon_vertices[:, 0], polygon_vertices[:, 1])
+    image[rr, cc] = 255
+
+
+if __name__ == "__main__":
+
+    db = DatabaseManager()
+    model_name = "c17"
+    model_id = db.get_models(filters={"model_name": model_name})[0]["id"]
+    print(f"Model name {model_name}, id {model_id}")
+    detections = db.get_detections(filters={"model_id": model_id})
+
+    file_stems = set()
+
+    for detection in detections:
+        file_stems.add(detection["image_filename"].split("_")[0])
+
+    print("Files:", file_stems)
+
+    for stem in file_stems:
+        print(stem)
+        detections = db.get_detections(filters={"model_id": model_id, "i.filename": f"LIKE %{stem}%"})
+        annotations = []
+        for detection in detections:
+            source_path = Path(detection["metadata"]["source_path"])
+            image, metadata = read_image(source_path)
+
+            offset = np.array(list(map(int, metadata["tile_section"].split(","))))[::-1]
+            scale = np.array(list(map(int, metadata["patch_size"].split(","))))[::-1]
+            # tile_size = np.array(list(map(int, metadata["tile_size"].split(","))))
+            segmentation = np.array(detection["segmentation_mask"])  # * tile_size
+
+            # print(source_path, image, metadata, segmentation.shape)
+            # print(offset)
+            # print(scale)
+            # print(segmentation)
+
+            # segmentation = (segmentation + offset * tile_size) / (tile_size * scale)
+            segmentation = (segmentation + offset) / scale
+
+            yolo_annotation = f"{detection['metadata']['class_id']} " + " ".join(
+                [f"{x:.6f} {y:.6f}" for x, y in segmentation]
+            )
+            annotations.append(yolo_annotation)
+            # print(segmentation)
+            # print(yolo_annotation)
+
+            # aa
+            print(
+                "  ",
+                detection["model_name"],
+                detection["image_id"],
+                detection["image_filename"],
+                source_path,
+                metadata["label_path"],
+            )
+            # section_i_section_j = detection["image_filename"].split("_")[1].split(".")[0]
+            # print("    ", section_i_section_j)
+
+        label_path = metadata["label_path"]
+        print("  ", label_path)
+        with open(label_path, "w") as f:
+            f.write("\n".join(annotations))
+
+    exit()
+
+    for detection in detections:
+        print(detection["model_name"], detection["image_id"], detection["image_filename"])
+
+    print(detections[0])
+    # polygon_vertices = np.array([[10, 10], [50, 10], [50, 50], [10, 50]])
+
+    # image = np.zeros((100, 100), dtype=np.uint8)
+
+    # rr, cc = polygon(polygon_vertices[:, 0], polygon_vertices[:, 1])
+
+    # image[rr, cc] = 255
+
+    # import matplotlib.pyplot as plt
+
+    # plt.imshow(image, cmap='gray')
+    # plt.show()

tests/show_yolo_seg.py

@@ -189,33 +189,38 @@ def main():
         # continue and just show image
     out = draw_annotations(img.copy(), labels, alpha=args.alpha, draw_bbox_for_poly=(not args.no_bbox))
 
-    lclass, coords = labels[0]
-    print(lclass, coords)
-    bbox = coords[:4]
-    print("bbox", bbox)
-    bbox = np.array(bbox) * np.array([img.shape[1], img.shape[0], img.shape[1], img.shape[0]])
-    yc, xc, h, w = bbox
-    print("bbox", bbox)
-
-    # polyline = np.array(coords[4:]).reshape(-1, 2) * np.array([img.shape[1], img.shape[0]])
-    polyline = np.array(coords).reshape(-1, 2) * np.array([img.shape[1], img.shape[0]])
-    print("pl", coords[4:])
-    print("pl", polyline)
-
-    # Convert BGR -> RGB for matplotlib display
-    # out_rgb = cv2.cvtColor(out, cv2.COLOR_BGR2RGB)
     out_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
-    # out_rgb = Image()
     plt.figure(figsize=(10, 10 * out.shape[0] / out.shape[1]))
-    plt.imshow(out_rgb)
-    plt.plot(polyline[:, 0], polyline[:, 1], "y", linewidth=2)
     if 0:
-        plt.plot(
+        plt.imshow(out_rgb.transpose(1, 0, 2))
-            [yc - h / 2, yc - h / 2, yc + h / 2, yc + h / 2, yc - h / 2],
+    else:
-            [xc - w / 2, xc + w / 2, xc + w / 2, xc - w / 2, xc - w / 2],
+        plt.imshow(out_rgb)
-            "r",
+
-            linewidth=2,
+    for label in labels:
-        )
+        lclass, coords = label
+        # print(lclass, coords)
+        bbox = coords[:4]
+        # print("bbox", bbox)
+        bbox = np.array(bbox) * np.array([img.shape[1], img.shape[0], img.shape[1], img.shape[0]])
+        yc, xc, h, w = bbox
+        # print("bbox", bbox)
+
+        # polyline = np.array(coords[4:]).reshape(-1, 2) * np.array([img.shape[1], img.shape[0]])
+        polyline = np.array(coords).reshape(-1, 2) * np.array([img.shape[1], img.shape[0]])
+        # print("pl", coords[4:])
+        # print("pl", polyline)
+
+        # Convert BGR -> RGB for matplotlib display
+        # out_rgb = cv2.cvtColor(out, cv2.COLOR_BGR2RGB)
+        # out_rgb = Image()
+        plt.plot(polyline[:, 0], polyline[:, 1], "y", linewidth=2)
+        if 0:
+            plt.plot(
+                [yc - h / 2, yc - h / 2, yc + h / 2, yc + h / 2, yc - h / 2],
+                [xc - w / 2, xc + w / 2, xc + w / 2, xc - w / 2, xc - w / 2],
+                "r",
+                linewidth=2,
+            )
 
     # plt.axis("off")
     plt.title(f"{img_path.name} ({lbl_path.name})")