Implement Wizard for Creating Classification Models (#20622)

* Implement extraction of images for classification state models

* Add object classification dataset preparation

* Add first step wizard

* Update i18n

* Add state classification image selection step

* Improve box handling

* Add object selector

* Improve object cropping implementation

* Fix state classification selection

* Finalize training and image selection step

* Cleanup

* Design optimizations

* Cleanup mobile styling

* Update no models screen

* Cleanups and fixes

* Fix bugs

* Improve model training and creation process

* Cleanup

* Dynamically add metrics for new model

* Add loading when hitting continue

* Improve image selection mechanism

* Remove unused translation keys

* Adjust wording

* Add retry button for image generation

* Make no models view more specific

* Adjust plus icon

* Adjust form label

* Start with correct type selected

* Cleanup sizing and more font colors

* Small tweaks

* Add tips and more info

* Cleanup dialog sizing

* Add cursor rule for frontend

* Cleanup

* remove underline

* Lazy loading

frigate/api/app.py

@@ -387,20 +387,28 @@ def config_set(request: Request, body: AppConfigSetBody):
         old_config: FrigateConfig = request.app.frigate_config
         request.app.frigate_config = config
 
-        if body.update_topic and body.update_topic.startswith("config/cameras/"):
-            _, _, camera, field = body.update_topic.split("/")
+        if body.update_topic:
+            if body.update_topic.startswith("config/cameras/"):
+                _, _, camera, field = body.update_topic.split("/")
 
-            if field == "add":
-                settings = config.cameras[camera]
-            elif field == "remove":
-                settings = old_config.cameras[camera]
+                if field == "add":
+                    settings = config.cameras[camera]
+                elif field == "remove":
+                    settings = old_config.cameras[camera]
+                else:
+                    settings = config.get_nested_object(body.update_topic)
+
+                request.app.config_publisher.publish_update(
+                    CameraConfigUpdateTopic(CameraConfigUpdateEnum[field], camera),
+                    settings,
+                )
             else:
+                # Handle nested config updates (e.g., config/classification/custom/{name})
                 settings = config.get_nested_object(body.update_topic)
-
-            request.app.config_publisher.publish_update(
-                CameraConfigUpdateTopic(CameraConfigUpdateEnum[field], camera),
-                settings,
-            )
+                if settings:
+                    request.app.config_publisher.publisher.publish(
+                        body.update_topic, settings
+                    )
 
     return JSONResponse(
         content=(

frigate/api/classification.py

@@ -3,7 +3,9 @@
 import datetime
 import logging
 import os
+import random
 import shutil
+import string
 from typing import Any
 
 import cv2
@@ -17,6 +19,8 @@ from frigate.api.auth import require_role
 from frigate.api.defs.request.classification_body import (
     AudioTranscriptionBody,
     DeleteFaceImagesBody,
+    GenerateObjectExamplesBody,
+    GenerateStateExamplesBody,
     RenameFaceBody,
 )
 from frigate.api.defs.response.classification_response import (
@@ -30,6 +34,10 @@ from frigate.config.camera import DetectConfig
 from frigate.const import CLIPS_DIR, FACE_DIR
 from frigate.embeddings import EmbeddingsContext
 from frigate.models import Event
+from frigate.util.classification import (
+    collect_object_classification_examples,
+    collect_state_classification_examples,
+)
 from frigate.util.path import get_event_snapshot
 
 logger = logging.getLogger(__name__)
@@ -159,8 +167,7 @@ def train_face(request: Request, name: str, body: dict = None):
     new_name = f"{sanitized_name}-{datetime.datetime.now().timestamp()}.webp"
     new_file_folder = os.path.join(FACE_DIR, f"{sanitized_name}")
 
-    if not os.path.exists(new_file_folder):
-        os.mkdir(new_file_folder)
+    os.makedirs(new_file_folder, exist_ok=True)
 
     if training_file_name:
         shutil.move(training_file, os.path.join(new_file_folder, new_name))
@@ -701,13 +708,14 @@ def categorize_classification_image(request: Request, name: str, body: dict = No
             status_code=404,
         )
 
-    new_name = f"{category}-{datetime.datetime.now().timestamp()}.png"
+    random_id = "".join(random.choices(string.ascii_lowercase + string.digits, k=6))
+    timestamp = datetime.datetime.now().timestamp()
+    new_name = f"{category}-{timestamp}-{random_id}.png"
     new_file_folder = os.path.join(
         CLIPS_DIR, sanitize_filename(name), "dataset", category
     )
 
-    if not os.path.exists(new_file_folder):
-        os.mkdir(new_file_folder)
+    os.makedirs(new_file_folder, exist_ok=True)
 
     # use opencv because webp images can not be used to train
     img = cv2.imread(training_file)
@@ -756,3 +764,43 @@ def delete_classification_train_images(request: Request, name: str, body: dict =
         content=({"success": True, "message": "Successfully deleted faces."}),
         status_code=200,
     )
+
+
+@router.post(
+    "/classification/generate_examples/state",
+    response_model=GenericResponse,
+    dependencies=[Depends(require_role(["admin"]))],
+    summary="Generate state classification examples",
+)
+async def generate_state_examples(request: Request, body: GenerateStateExamplesBody):
+    """Generate examples for state classification."""
+    model_name = sanitize_filename(body.model_name)
+    cameras_normalized = {
+        camera_name: tuple(crop)
+        for camera_name, crop in body.cameras.items()
+        if camera_name in request.app.frigate_config.cameras
+    }
+
+    collect_state_classification_examples(model_name, cameras_normalized)
+
+    return JSONResponse(
+        content={"success": True, "message": "Example generation completed"},
+        status_code=200,
+    )
+
+
+@router.post(
+    "/classification/generate_examples/object",
+    response_model=GenericResponse,
+    dependencies=[Depends(require_role(["admin"]))],
+    summary="Generate object classification examples",
+)
+async def generate_object_examples(request: Request, body: GenerateObjectExamplesBody):
+    """Generate examples for object classification."""
+    model_name = sanitize_filename(body.model_name)
+    collect_object_classification_examples(model_name, body.label)
+
+    return JSONResponse(
+        content={"success": True, "message": "Example generation completed"},
+        status_code=200,
+    )

frigate/api/defs/request/classification_body.py

@@ -1,17 +1,31 @@
-from typing import List
+from typing import Dict, List, Tuple
 
 from pydantic import BaseModel, Field
 
 
 class RenameFaceBody(BaseModel):
-    new_name: str
+    new_name: str = Field(description="New name for the face")
 
 
 class AudioTranscriptionBody(BaseModel):
-    event_id: str
+    event_id: str = Field(description="ID of the event to transcribe audio for")
 
 
 class DeleteFaceImagesBody(BaseModel):
     ids: List[str] = Field(
         description="List of image filenames to delete from the face folder"
     )
+
+
+class GenerateStateExamplesBody(BaseModel):
+    model_name: str = Field(description="Name of the classification model")
+    cameras: Dict[str, Tuple[float, float, float, float]] = Field(
+        description="Dictionary mapping camera names to normalized crop coordinates in [x1, y1, x2, y2] format (values 0-1)"
+    )
+
+
+class GenerateObjectExamplesBody(BaseModel):
+    model_name: str = Field(description="Name of the classification model")
+    label: str = Field(
+        description="Object label to collect examples for (e.g., 'person', 'car')"
+    )