Still create all classes

We stil need to create all classes even if the user didn't assign images to them.

frigate/api/classification.py

@@ -870,6 +870,46 @@ def categorize_classification_image(request: Request, name: str, body: dict = No
     )
 
 
+@router.post(
+    "/classification/{name}/dataset/{category}/create",
+    response_model=GenericResponse,
+    dependencies=[Depends(require_role(["admin"]))],
+    summary="Create an empty classification category folder",
+    description="""Creates an empty folder for a classification category.
+    This is used to create folders for categories that don't have images yet.
+    Returns a success message or an error if the name is invalid.""",
+)
+def create_classification_category(request: Request, name: str, category: str):
+    config: FrigateConfig = request.app.frigate_config
+
+    if name not in config.classification.custom:
+        return JSONResponse(
+            content=(
+                {
+                    "success": False,
+                    "message": f"{name} is not a known classification model.",
+                }
+            ),
+            status_code=404,
+        )
+
+    category_folder = os.path.join(
+        CLIPS_DIR, sanitize_filename(name), "dataset", sanitize_filename(category)
+    )
+
+    os.makedirs(category_folder, exist_ok=True)
+
+    return JSONResponse(
+        content=(
+            {
+                "success": True,
+                "message": f"Successfully created category folder: {category}",
+            }
+        ),
+        status_code=200,
+    )
+
+
 @router.post(
     "/classification/{name}/train/delete",
     response_model=GenericResponse,

web/public/locales/en/views/classificationModel.json

@@ -166,6 +166,7 @@
       "noImages": "No sample images generated",
       "classifying": "Classifying & Training...",
       "trainingStarted": "Training started successfully",
+      "modelCreated": "Model created successfully. Use the Recent Classifications view to add images for missing states, then train the model.",
       "errors": {
         "noCameras": "No cameras configured",
         "noObjectLabel": "No object label selected",

web/src/components/classification/wizard/Step3ChooseExamples.tsx

@@ -141,15 +141,49 @@ export default function Step3ChooseExamples({
       );
       await Promise.all(categorizePromises);
 
-      // Step 3: Kick off training
-      await axios.post(`/classification/${step1Data.modelName}/train`);
+      // Step 2.5: Create empty folders for classes that don't have any images
+      // This ensures all classes are available in the dataset view later
+      const classesWithImages = new Set(
+        Object.values(classifications).filter((c) => c && c !== "none"),
+      );
+      const emptyFolderPromises = step1Data.classes
+        .filter((className) => !classesWithImages.has(className))
+        .map((className) =>
+          axios.post(
+            `/classification/${step1Data.modelName}/dataset/${className}/create`,
+          ),
+        );
+      await Promise.all(emptyFolderPromises);
 
-      toast.success(t("wizard.step3.trainingStarted"), {
-        closeButton: true,
-      });
-      setIsTraining(true);
+      // Step 3: Determine if we should train
+      // For state models, we need ALL states to have examples
+      // For object models, we need at least 2 classes with images
+      const allStatesHaveExamplesForTraining =
+        step1Data.modelType !== "state" ||
+        step1Data.classes.every((className) =>
+          classesWithImages.has(className),
+        );
+      const shouldTrain =
+        allStatesHaveExamplesForTraining && classesWithImages.size >= 2;
+
+      // Step 4: Kick off training only if we have enough classes with images
+      if (shouldTrain) {
+        await axios.post(`/classification/${step1Data.modelName}/train`);
+
+        toast.success(t("wizard.step3.trainingStarted"), {
+          closeButton: true,
+        });
+        setIsTraining(true);
+      } else {
+        // Don't train - not all states have examples
+        toast.success(t("wizard.step3.modelCreated"), {
+          closeButton: true,
+        });
+        setIsTraining(false);
+        onClose();
+      }
     },
-    [step1Data, step2Data, t],
+    [step1Data, step2Data, t, onClose],
   );
 
   const handleContinueClassification = useCallback(async () => {