Miscellaneous Fixes (#20866)

* Don't warn when event ids have expired for trigger sync

* Import faster_whisper conditinally to avoid illegal instruction

* Catch OpenVINO runtime error

* fix race condition in detail stream context

navigating between tracked objects in Explore would sometimes prevent the object track from appearing

* Handle case where classification images are deleted

* Adjust default rounded corners on larger screens

* Improve flow handling for classification state

* Remove images when wizard is cancelled

* Improve deletion handling for classes

* Set constraints on review buffers

* Update to support correct data format

* Set minimum duration for recording based review items

* Use friendly name in review genai prompt

---------

Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>

frigate/api/classification.py

@@ -595,9 +595,13 @@ def get_classification_dataset(name: str):
             "last_training_image_count": 0,
             "current_image_count": current_image_count,
             "new_images_count": current_image_count,
+            "dataset_changed": current_image_count > 0,
         }
     else:
         last_training_count = metadata.get("last_training_image_count", 0)
+        # Dataset has changed if count is different (either added or deleted images)
+        dataset_changed = current_image_count != last_training_count
+        # Only show positive count for new images (ignore deletions in the count display)
         new_images_count = max(0, current_image_count - last_training_count)
         training_metadata = {
             "has_trained": True,
@@ -605,6 +609,7 @@ def get_classification_dataset(name: str):
             "last_training_image_count": last_training_count,
             "current_image_count": current_image_count,
             "new_images_count": new_images_count,
+            "dataset_changed": dataset_changed,
         }
 
     return JSONResponse(
@@ -948,31 +953,29 @@ async def generate_object_examples(request: Request, body: GenerateObjectExample
     dependencies=[Depends(require_role(["admin"]))],
     summary="Delete a classification model",
     description="""Deletes a specific classification model and all its associated data.
-    The name must exist in the classification models. Returns a success message or an error if the name is invalid.""",
+    Works even if the model is not in the config (e.g., partially created during wizard).
+    Returns a success message.""",
 )
 def delete_classification_model(request: Request, name: str):
-    config: FrigateConfig = request.app.frigate_config
+    sanitized_name = sanitize_filename(name)
-
-    if name not in config.classification.custom:
-        return JSONResponse(
-            content=(
-                {
-                    "success": False,
-                    "message": f"{name} is not a known classification model.",
-                }
-            ),
-            status_code=404,
-        )
 
     # Delete the classification model's data directory in clips
-    data_dir = os.path.join(CLIPS_DIR, sanitize_filename(name))
+    data_dir = os.path.join(CLIPS_DIR, sanitized_name)
     if os.path.exists(data_dir):
-        shutil.rmtree(data_dir)
+        try:
+            shutil.rmtree(data_dir)
+            logger.info(f"Deleted classification data directory for {name}")
+        except Exception as e:
+            logger.debug(f"Failed to delete data directory for {name}: {e}")
 
     # Delete the classification model's files in model_cache
-    model_dir = os.path.join(MODEL_CACHE_DIR, sanitize_filename(name))
+    model_dir = os.path.join(MODEL_CACHE_DIR, sanitized_name)
     if os.path.exists(model_dir):
-        shutil.rmtree(model_dir)
+        try:
+            shutil.rmtree(model_dir)
+            logger.info(f"Deleted classification model directory for {name}")
+        except Exception as e:
+            logger.debug(f"Failed to delete model directory for {name}: {e}")
 
     return JSONResponse(
         content=(

frigate/config/camera/camera.py

@@ -177,6 +177,12 @@ class CameraConfig(FrigateBaseModel):
     def ffmpeg_cmds(self) -> list[dict[str, list[str]]]:
         return self._ffmpeg_cmds
 
+    def get_formatted_name(self) -> str:
+        """Return the friendly name if set, otherwise return a formatted version of the camera name."""
+        if self.friendly_name:
+            return self.friendly_name
+        return self.name.replace("_", " ").title() if self.name else ""
+
     def create_ffmpeg_cmds(self):
         if "_ffmpeg_cmds" in self:
             return

frigate/config/camera/zone.py

@@ -56,6 +56,12 @@ class ZoneConfig(BaseModel):
     def contour(self) -> np.ndarray:
         return self._contour
 
+    def get_formatted_name(self, zone_name: str) -> str:
+        """Return the friendly name if set, otherwise return a formatted version of the zone name."""
+        if self.friendly_name:
+            return self.friendly_name
+        return zone_name.replace("_", " ").title()
+
     @field_validator("objects", mode="before")
     @classmethod
     def validate_objects(cls, v):

frigate/data_processing/common/audio_transcription/model.py

@@ -4,7 +4,6 @@ import logging
 import os
 
 import sherpa_onnx
-from faster_whisper.utils import download_model
 
 from frigate.comms.inter_process import InterProcessRequestor
 from frigate.const import MODEL_CACHE_DIR
@@ -25,6 +24,9 @@ class AudioTranscriptionModelRunner:
 
         if model_size == "large":
             # use the Whisper download function instead of our own
+            # Import dynamically to avoid crashes on systems without AVX support
+            from faster_whisper.utils import download_model
+
             logger.debug("Downloading Whisper audio transcription model")
             download_model(
                 size_or_id="small" if device == "cuda" else "tiny",

frigate/data_processing/post/audio_transcription.py

@@ -6,7 +6,6 @@ import threading
 import time
 from typing import Optional
 
-from faster_whisper import WhisperModel
 from peewee import DoesNotExist
 
 from frigate.comms.inter_process import InterProcessRequestor
@@ -51,6 +50,9 @@ class AudioTranscriptionPostProcessor(PostProcessorApi):
 
     def __build_recognizer(self) -> None:
         try:
+            # Import dynamically to avoid crashes on systems without AVX support
+            from faster_whisper import WhisperModel
+
             self.recognizer = WhisperModel(
                 model_size_or_path="small",
                 device="cuda"

frigate/data_processing/post/review_descriptions.py

@@ -16,6 +16,7 @@ from peewee import DoesNotExist
 from frigate.comms.embeddings_updater import EmbeddingsRequestEnum
 from frigate.comms.inter_process import InterProcessRequestor
 from frigate.config import FrigateConfig
+from frigate.config.camera import CameraConfig
 from frigate.config.camera.review import GenAIReviewConfig, ImageSourceEnum
 from frigate.const import CACHE_DIR, CLIPS_DIR, UPDATE_REVIEW_DESCRIPTION
 from frigate.data_processing.types import PostProcessDataEnum
@@ -30,6 +31,7 @@ from ..types import DataProcessorMetrics
 logger = logging.getLogger(__name__)
 
 RECORDING_BUFFER_EXTENSION_PERCENT = 0.10
+MIN_RECORDING_DURATION = 10
 
 
 class ReviewDescriptionProcessor(PostProcessorApi):
@@ -130,7 +132,17 @@ class ReviewDescriptionProcessor(PostProcessorApi):
 
             if image_source == ImageSourceEnum.recordings:
                 duration = final_data["end_time"] - final_data["start_time"]
-                buffer_extension = duration * RECORDING_BUFFER_EXTENSION_PERCENT
+                buffer_extension = min(
+                    10, max(2, duration * RECORDING_BUFFER_EXTENSION_PERCENT)
+                )
+
+                # Ensure minimum total duration for short review items
+                # This provides better context for brief events
+                total_duration = duration + (2 * buffer_extension)
+                if total_duration < MIN_RECORDING_DURATION:
+                    # Expand buffer to reach minimum duration, still respecting max of 10s per side
+                    additional_buffer_per_side = (MIN_RECORDING_DURATION - duration) / 2
+                    buffer_extension = min(10, additional_buffer_per_side)
 
                 thumbs = self.get_recording_frames(
                     camera,
@@ -182,7 +194,7 @@ class ReviewDescriptionProcessor(PostProcessorApi):
                     self.requestor,
                     self.genai_client,
                     self.review_desc_speed,
-                    camera,
+                    camera_config,
                     final_data,
                     thumbs,
                     camera_config.review.genai,
@@ -411,7 +423,7 @@ def run_analysis(
     requestor: InterProcessRequestor,
     genai_client: GenAIClient,
     review_inference_speed: InferenceSpeed,
-    camera: str,
+    camera_config: CameraConfig,
     final_data: dict[str, str],
     thumbs: list[bytes],
     genai_config: GenAIReviewConfig,
@@ -419,10 +431,19 @@ def run_analysis(
     attribute_labels: list[str],
 ) -> None:
     start = datetime.datetime.now().timestamp()
+
+    # Format zone names using zone config friendly names if available
+    formatted_zones = []
+    for zone_name in final_data["data"]["zones"]:
+        if zone_name in camera_config.zones:
+            formatted_zones.append(
+                camera_config.zones[zone_name].get_formatted_name(zone_name)
+            )
+
     analytics_data = {
         "id": final_data["id"],
-        "camera": camera,
+        "camera": camera_config.get_formatted_name(),
-        "zones": final_data["data"]["zones"],
+        "zones": formatted_zones,
         "start": datetime.datetime.fromtimestamp(final_data["start_time"]).strftime(
             "%A, %I:%M %p"
         ),

frigate/detectors/detection_runners.py

@@ -394,7 +394,11 @@ class OpenVINOModelRunner(BaseModelRunner):
                     self.infer_request.set_input_tensor(input_index, input_tensor)
 
                 # Run inference
-                self.infer_request.infer()
+                try:
+                    self.infer_request.infer()
+                except Exception as e:
+                    logger.error(f"Error during OpenVINO inference: {e}")
+                    return []
 
             # Get all output tensors
             outputs = []

frigate/embeddings/embeddings.py

@@ -472,7 +472,7 @@ class Embeddings:
                                 )
                                 thumbnail_missing = True
                         except DoesNotExist:
-                            logger.warning(
+                            logger.debug(
                                 f"Event ID {trigger.data} for trigger {trigger_name} does not exist."
                             )
                             continue

frigate/genai/__init__.py

@@ -51,8 +51,7 @@ class GenAIClient:
         def get_concern_prompt() -> str:
             if concerns:
                 concern_list = "\n    - ".join(concerns)
-                return f"""
+                return f"""- `other_concerns` (list of strings): Include a list of any of the following concerns that are occurring:
-- `other_concerns` (list of strings): Include a list of any of the following concerns that are occurring:
     - {concern_list}"""
             else:
                 return ""
@@ -70,7 +69,7 @@ class GenAIClient:
                 return "\n- (No objects detected)"
 
         context_prompt = f"""
-Your task is to analyze the sequence of images ({len(thumbnails)} total) taken in chronological order from the perspective of the {review_data["camera"].replace("_", " ")} security camera.
+Your task is to analyze the sequence of images ({len(thumbnails)} total) taken in chronological order from the perspective of the {review_data["camera"]} security camera.
 
 ## Normal Activity Patterns for This Property
 
@@ -110,7 +109,7 @@ Your response MUST be a flat JSON object with:
 
 - Frame 1 = earliest, Frame {len(thumbnails)} = latest
 - Activity started at {review_data["start"]} and lasted {review_data["duration"]} seconds
-- Zones involved: {", ".join(z.replace("_", " ").title() for z in review_data["zones"]) or "None"}
+- Zones involved: {", ".join(review_data["zones"]) if review_data["zones"] else "None"}
 
 ## Objects in Scene
 

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

@@ -16,6 +16,7 @@
   "tooltip": {
     "trainingInProgress": "Model is currently training",
     "noNewImages": "No new images to train. Classify more images in the dataset first.",
+    "noChanges": "No changes to the dataset since last training.",
     "modelNotReady": "Model is not ready for training"
   },
   "toast": {
@@ -43,7 +44,9 @@
   },
   "deleteCategory": {
     "title": "Delete Class",
-    "desc": "Are you sure you want to delete the class {{name}}? This will permanently delete all associated images and require re-training the model."
+    "desc": "Are you sure you want to delete the class {{name}}? This will permanently delete all associated images and require re-training the model.",
+    "minClassesTitle": "Cannot Delete Class",
+    "minClassesDesc": "A classification model must have at least 2 classes. Add another class before deleting this one."
   },
   "deleteModel": {
     "title": "Delete Classification Model",

web/src/components/classification/ClassificationModelEditDialog.tsx

@@ -28,6 +28,7 @@ import {
   CustomClassificationModelConfig,
   FrigateConfig,
 } from "@/types/frigateConfig";
+import { ClassificationDatasetResponse } from "@/types/classification";
 import { getTranslatedLabel } from "@/utils/i18n";
 import { zodResolver } from "@hookform/resolvers/zod";
 import axios from "axios";
@@ -140,16 +141,19 @@ export default function ClassificationModelEditDialog({
   });
 
   // Fetch dataset to get current classes for state models
-  const { data: dataset } = useSWR<{
+  const { data: dataset } = useSWR<ClassificationDatasetResponse>(
-    [id: string]: string[];
+    isStateModel ? `classification/${model.name}/dataset` : null,
-  }>(isStateModel ? `classification/${model.name}/dataset` : null, {
+    {
-    revalidateOnFocus: false,
+      revalidateOnFocus: false,
-  });
+    },
+  );
 
   // Update form with classes from dataset when loaded
   useEffect(() => {
-    if (isStateModel && dataset) {
+    if (isStateModel && dataset?.categories) {
-      const classes = Object.keys(dataset).filter((key) => key !== "none");
+      const classes = Object.keys(dataset.categories).filter(
+        (key) => key !== "none",
+      );
       if (classes.length > 0) {
         (form as ReturnType<typeof useForm<StateFormData>>).setValue(
           "classes",

web/src/components/classification/ClassificationModelWizardDialog.tsx

@@ -15,6 +15,7 @@ import Step3ChooseExamples, {
 } from "./wizard/Step3ChooseExamples";
 import { cn } from "@/lib/utils";
 import { isDesktop } from "react-device-detect";
+import axios from "axios";
 
 const OBJECT_STEPS = [
   "wizard.steps.nameAndDefine",
@@ -120,7 +121,18 @@ export default function ClassificationModelWizardDialog({
     dispatch({ type: "PREVIOUS_STEP" });
   };
 
-  const handleCancel = () => {
+  const handleCancel = async () => {
+    // Clean up any generated training images if we're cancelling from Step 3
+    if (wizardState.step1Data && wizardState.step3Data?.examplesGenerated) {
+      try {
+        await axios.delete(
+          `/classification/${wizardState.step1Data.modelName}`,
+        );
+      } catch (error) {
+        // Silently fail - user is already cancelling
+      }
+    }
+
     dispatch({ type: "RESET" });
     onClose();
   };

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

@@ -165,18 +165,15 @@ export default function Step3ChooseExamples({
     const isLastClass = currentClassIndex === allClasses.length - 1;
 
     if (isLastClass) {
-      // Assign remaining unclassified images
+      // For object models, assign remaining unclassified images to "none"
-      unknownImages.slice(0, 24).forEach((imageName) => {
+      // For state models, this should never happen since we require all images to be classified
-        if (!newClassifications[imageName]) {
+      if (step1Data.modelType !== "state") {
-          // For state models with 2 classes, assign to the last class
+        unknownImages.slice(0, 24).forEach((imageName) => {
-          // For object models, assign to "none"
+          if (!newClassifications[imageName]) {
-          if (step1Data.modelType === "state" && allClasses.length === 2) {
-            newClassifications[imageName] = allClasses[allClasses.length - 1];
-          } else {
             newClassifications[imageName] = "none";
           }
-        }
+        });
-      });
+      }
 
       // All done, trigger training immediately
       setImageClassifications(newClassifications);
@@ -316,8 +313,15 @@ export default function Step3ChooseExamples({
       return images;
     }
 
-    return images.filter((img) => !imageClassifications[img]);
+    // If we're viewing a previous class (going back), show images for that class
-  }, [unknownImages, imageClassifications]);
+    // Otherwise show only unclassified images
+    const currentClassInView = allClasses[currentClassIndex];
+    return images.filter((img) => {
+      const imgClass = imageClassifications[img];
+      // Show if: unclassified OR classified with current class we're viewing
+      return !imgClass || imgClass === currentClassInView;
+    });
+  }, [unknownImages, imageClassifications, allClasses, currentClassIndex]);
 
   const allImagesClassified = useMemo(() => {
     return unclassifiedImages.length === 0;
@@ -326,15 +330,26 @@ export default function Step3ChooseExamples({
   // For state models on the last class, require all images to be classified
   const isLastClass = currentClassIndex === allClasses.length - 1;
   const canProceed = useMemo(() => {
-    if (
+    if (step1Data.modelType === "state" && isLastClass) {
-      step1Data.modelType === "state" &&
+      // Check if all 24 images will be classified after current selections are applied
-      isLastClass &&
+      const totalImages = unknownImages.slice(0, 24).length;
-      !allImagesClassified
+
-    ) {
+      // Count images that will be classified (either already classified or currently selected)
-      return false;
+      const allImages = unknownImages.slice(0, 24);
+      const willBeClassified = allImages.filter((img) => {
+        return imageClassifications[img] || selectedImages.has(img);
+      }).length;
+
+      return willBeClassified >= totalImages;
     }
     return true;
-  }, [step1Data.modelType, isLastClass, allImagesClassified]);
+  }, [
+    step1Data.modelType,
+    isLastClass,
+    unknownImages,
+    imageClassifications,
+    selectedImages,
+  ]);
 
   const handleBack = useCallback(() => {
     if (currentClassIndex > 0) {

web/src/components/overlay/ImageShadowOverlay.tsx

@@ -12,13 +12,13 @@ export function ImageShadowOverlay({
     <>
       <div
         className={cn(
-          "pointer-events-none absolute inset-x-0 top-0 z-10 h-[30%] w-full rounded-lg bg-gradient-to-b from-black/20 to-transparent md:rounded-2xl",
+          "pointer-events-none absolute inset-x-0 top-0 z-10 h-[30%] w-full rounded-lg bg-gradient-to-b from-black/20 to-transparent",
           upperClassName,
         )}
       />
       <div
         className={cn(
-          "pointer-events-none absolute inset-x-0 bottom-0 z-10 h-[10%] w-full rounded-lg bg-gradient-to-t from-black/20 to-transparent md:rounded-2xl",
+          "pointer-events-none absolute inset-x-0 bottom-0 z-10 h-[10%] w-full rounded-lg bg-gradient-to-t from-black/20 to-transparent",
           lowerClassName,
         )}
       />

web/src/components/player/BirdseyeLivePlayer.tsx

@@ -77,7 +77,10 @@ export default function BirdseyeLivePlayer({
       )}
       onClick={onClick}
     >
-      <ImageShadowOverlay />
+      <ImageShadowOverlay
+        upperClassName="md:rounded-2xl"
+        lowerClassName="md:rounded-2xl"
+      />
       <div className="size-full" ref={playerRef}>
         {player}
       </div>

web/src/components/player/LivePlayer.tsx

@@ -331,7 +331,10 @@ export default function LivePlayer({
     >
       {cameraEnabled &&
         ((showStillWithoutActivity && !liveReady) || liveReady) && (
-          <ImageShadowOverlay />
+          <ImageShadowOverlay
+            upperClassName="md:rounded-2xl"
+            lowerClassName="md:rounded-2xl"
+          />
         )}
       {player}
       {cameraEnabled &&

web/src/context/detail-stream-context.tsx

@@ -1,4 +1,10 @@
-import React, { createContext, useContext, useState, useEffect } from "react";
+import React, {
+  createContext,
+  useContext,
+  useState,
+  useEffect,
+  useRef,
+} from "react";
 import { FrigateConfig } from "@/types/frigateConfig";
 import useSWR from "swr";
 
@@ -36,6 +42,23 @@ export function DetailStreamProvider({
     () => initialSelectedObjectIds ?? [],
   );
 
+  // When the parent provides a new initialSelectedObjectIds (for example
+  // when navigating between search results) update the selection so children
+  // like `ObjectTrackOverlay` receive the new ids immediately. We only
+  // perform this update when the incoming value actually changes.
+  useEffect(() => {
+    if (
+      initialSelectedObjectIds &&
+      (initialSelectedObjectIds.length !== selectedObjectIds.length ||
+        initialSelectedObjectIds.some((v, i) => selectedObjectIds[i] !== v))
+    ) {
+      setSelectedObjectIds(initialSelectedObjectIds);
+    }
+    // Intentionally include selectedObjectIds to compare previous value and
+    // avoid overwriting user interactions unless the incoming prop changed.
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [initialSelectedObjectIds]);
+
   const toggleObjectSelection = (id: string | undefined) => {
     if (id === undefined) {
       setSelectedObjectIds([]);
@@ -63,10 +86,33 @@ export function DetailStreamProvider({
     setAnnotationOffset(cfgOffset);
   }, [config, camera]);
 
-  // Clear selected objects when exiting detail mode or changing cameras
+  // Clear selected objects when exiting detail mode or when the camera
+  // changes for providers that are not initialized with an explicit
+  // `initialSelectedObjectIds` (e.g., the RecordingView). For providers
+  // that receive `initialSelectedObjectIds` (like SearchDetailDialog) we
+  // avoid clearing on camera change to prevent a race with children that
+  // immediately set selection when mounting.
+  const prevCameraRef = useRef<string | undefined>(undefined);
   useEffect(() => {
-    setSelectedObjectIds([]);
+    // Always clear when leaving detail mode
-  }, [isDetailMode, camera]);
+    if (!isDetailMode) {
+      setSelectedObjectIds([]);
+      prevCameraRef.current = camera;
+      return;
+    }
+
+    // If camera changed and the parent did not provide initialSelectedObjectIds,
+    // clear selection to preserve previous behavior.
+    if (
+      prevCameraRef.current !== undefined &&
+      prevCameraRef.current !== camera &&
+      initialSelectedObjectIds === undefined
+    ) {
+      setSelectedObjectIds([]);
+    }
+
+    prevCameraRef.current = camera;
+  }, [isDetailMode, camera, initialSelectedObjectIds]);
 
   const value: DetailStreamContextType = {
     selectedObjectIds,

web/src/types/classification.ts

@@ -20,3 +20,17 @@ export type ClassificationThreshold = {
   recognition: number;
   unknown: number;
 };
+
+export type ClassificationDatasetResponse = {
+  categories: {
+    [id: string]: string[];
+  };
+  training_metadata: {
+    has_trained: boolean;
+    last_training_date: string | null;
+    last_training_image_count: number;
+    current_image_count: number;
+    new_images_count: number;
+    dataset_changed: boolean;
+  } | null;
+};

web/src/views/classification/ModelSelectionView.tsx

@@ -11,6 +11,7 @@ import {
   CustomClassificationModelConfig,
   FrigateConfig,
 } from "@/types/frigateConfig";
+import { ClassificationDatasetResponse } from "@/types/classification";
 import { useCallback, useEffect, useMemo, useState } from "react";
 import { useTranslation } from "react-i18next";
 import { FaFolderPlus } from "react-icons/fa";
@@ -209,9 +210,10 @@ type ModelCardProps = {
 function ModelCard({ config, onClick, onUpdate, onDelete }: ModelCardProps) {
   const { t } = useTranslation(["views/classificationModel"]);
 
-  const { data: dataset } = useSWR<{
+  const { data: dataset } = useSWR<ClassificationDatasetResponse>(
-    [id: string]: string[];
+    `classification/${config.name}/dataset`,
-  }>(`classification/${config.name}/dataset`, { revalidateOnFocus: false });
+    { revalidateOnFocus: false },
+  );
 
   const [deleteDialogOpen, setDeleteDialogOpen] = useState(false);
   const [editDialogOpen, setEditDialogOpen] = useState(false);
@@ -260,20 +262,25 @@ function ModelCard({ config, onClick, onUpdate, onDelete }: ModelCardProps) {
   }, []);
 
   const coverImage = useMemo(() => {
-    if (!dataset) {
+    if (!dataset || !dataset.categories) {
       return undefined;
     }
 
-    const keys = Object.keys(dataset).filter((key) => key != "none");
+    const keys = Object.keys(dataset.categories).filter((key) => key != "none");
-    const selectedKey = keys[0];
+    if (keys.length === 0) {
+      return undefined;
+    }
 
-    if (!dataset[selectedKey]) {
+    const selectedKey = keys[0];
+    const images = dataset.categories[selectedKey];
+
+    if (!images || images.length === 0) {
       return undefined;
     }
 
     return {
       name: selectedKey,
-      img: dataset[selectedKey][0],
+      img: images[0],
     };
   }, [dataset]);
 
@@ -317,11 +324,19 @@ function ModelCard({ config, onClick, onUpdate, onDelete }: ModelCardProps) {
         )}
         onClick={onClick}
       >
-        <img
+        {coverImage ? (
-          className="size-full"
+          <>
-          src={`${baseUrl}clips/${config.name}/dataset/${coverImage?.name}/${coverImage?.img}`}
+            <img
-        />
+              className="size-full"
-        <ImageShadowOverlay lowerClassName="h-[30%] z-0" />
+              src={`${baseUrl}clips/${config.name}/dataset/${coverImage.name}/${coverImage.img}`}
+            />
+            <ImageShadowOverlay lowerClassName="h-[30%] z-0" />
+          </>
+        ) : (
+          <div className="flex size-full items-center justify-center bg-background_alt">
+            <MdModelTraining className="size-16 text-muted-foreground" />
+          </div>
+        )}
         <div className="absolute bottom-2 left-3 text-lg text-white smart-capitalize">
           {config.name}
         </div>

web/src/views/classification/ModelTrainingView.tsx

@@ -59,7 +59,11 @@ import { useNavigate } from "react-router-dom";
 import { IoMdArrowRoundBack } from "react-icons/io";
 import TrainFilterDialog from "@/components/overlay/dialog/TrainFilterDialog";
 import useApiFilter from "@/hooks/use-api-filter";
-import { ClassificationItemData, TrainFilter } from "@/types/classification";
+import {
+  ClassificationDatasetResponse,
+  ClassificationItemData,
+  TrainFilter,
+} from "@/types/classification";
 import {
   ClassificationCard,
   GroupedClassificationCard,
@@ -118,16 +122,10 @@ export default function ModelTrainingView({ model }: ModelTrainingViewProps) {
   const { data: trainImages, mutate: refreshTrain } = useSWR<string[]>(
     `classification/${model.name}/train`,
   );
-  const { data: datasetResponse, mutate: refreshDataset } = useSWR<{
+  const { data: datasetResponse, mutate: refreshDataset } =
-    categories: { [id: string]: string[] };
+    useSWR<ClassificationDatasetResponse>(
-    training_metadata: {
+      `classification/${model.name}/dataset`,
-      has_trained: boolean;
+    );
-      last_training_date: string | null;
-      last_training_image_count: number;
-      current_image_count: number;
-      new_images_count: number;
-    } | null;
-  }>(`classification/${model.name}/dataset`);
 
   const dataset = datasetResponse?.categories || {};
   const trainingMetadata = datasetResponse?.training_metadata;
@@ -264,10 +262,11 @@ export default function ModelTrainingView({ model }: ModelTrainingViewProps) {
               );
             }
 
+            // Always refresh dataset to update the categories list
+            refreshDataset();
+
             if (pageToggle == "train") {
               refreshTrain();
-            } else {
-              refreshDataset();
             }
           }
         })
@@ -445,7 +444,7 @@ export default function ModelTrainingView({ model }: ModelTrainingViewProps) {
                   variant={modelState == "failed" ? "destructive" : "select"}
                   disabled={
                     (modelState != "complete" && modelState != "failed") ||
-                    (trainingMetadata?.new_images_count ?? 0) === 0
+                    !trainingMetadata?.dataset_changed
                   }
                 >
                   {modelState == "training" ? (
@@ -466,14 +465,14 @@ export default function ModelTrainingView({ model }: ModelTrainingViewProps) {
                   )}
                 </Button>
               </TooltipTrigger>
-              {((trainingMetadata?.new_images_count ?? 0) === 0 ||
+              {(!trainingMetadata?.dataset_changed ||
                 (modelState != "complete" && modelState != "failed")) && (
                 <TooltipPortal>
                   <TooltipContent>
                     {modelState == "training"
                       ? t("tooltip.trainingInProgress")
-                      : trainingMetadata?.new_images_count === 0
+                      : !trainingMetadata?.dataset_changed
-                        ? t("tooltip.noNewImages")
+                        ? t("tooltip.noChanges")
                         : t("tooltip.modelNotReady")}
                   </TooltipContent>
                 </TooltipPortal>
@@ -571,27 +570,44 @@ function LibrarySelector({
       >
         <DialogContent>
           <DialogHeader>
-            <DialogTitle>{t("deleteCategory.title")}</DialogTitle>
+            <DialogTitle>
+              {Object.keys(dataset).length <= 2
+                ? t("deleteCategory.minClassesTitle")
+                : t("deleteCategory.title")}
+            </DialogTitle>
             <DialogDescription>
-              {t("deleteCategory.desc", { name: confirmDelete })}
+              {Object.keys(dataset).length <= 2
+                ? t("deleteCategory.minClassesDesc")
+                : t("deleteCategory.desc", { name: confirmDelete })}
             </DialogDescription>
           </DialogHeader>
           <div className="flex justify-end gap-2">
-            <Button variant="outline" onClick={() => setConfirmDelete(null)}>
+            {Object.keys(dataset).length <= 2 ? (
-              {t("button.cancel", { ns: "common" })}
+              <Button variant="outline" onClick={() => setConfirmDelete(null)}>
-            </Button>
+                {t("button.ok", { ns: "common" })}
-            <Button
+              </Button>
-              variant="destructive"
+            ) : (
-              className="text-white"
+              <>
-              onClick={() => {
+                <Button
-                if (confirmDelete) {
+                  variant="outline"
-                  handleDeleteCategory(confirmDelete);
+                  onClick={() => setConfirmDelete(null)}
-                  setConfirmDelete(null);
+                >
-                }
+                  {t("button.cancel", { ns: "common" })}
-              }}
+                </Button>
-            >
+                <Button
-              {t("button.delete", { ns: "common" })}
+                  variant="destructive"
-            </Button>
+                  className="text-white"
+                  onClick={() => {
+                    if (confirmDelete) {
+                      handleDeleteCategory(confirmDelete);
+                      setConfirmDelete(null);
+                    }
+                  }}
+                >
+                  {t("button.delete", { ns: "common" })}
+                </Button>
+              </>
+            )}
           </div>
         </DialogContent>
       </Dialog>