rename variable

mobile/openapi/README.md

@@ -96,6 +96,7 @@ Class | Method | HTTP request | Description
 *AssetApi* | [**deleteAssets**](doc//AssetApi.md#deleteassets) | **DELETE** /asset | 
 *AssetApi* | [**getAllAssets**](doc//AssetApi.md#getallassets) | **GET** /asset | 
 *AssetApi* | [**getAllUserAssetsByDeviceId**](doc//AssetApi.md#getalluserassetsbydeviceid) | **GET** /asset/device/{deviceId} | 
+*AssetApi* | [**getAssetDuplicates**](doc//AssetApi.md#getassetduplicates) | **GET** /asset/duplicates | 
 *AssetApi* | [**getAssetInfo**](doc//AssetApi.md#getassetinfo) | **GET** /asset/{id} | 
 *AssetApi* | [**getAssetSearchTerms**](doc//AssetApi.md#getassetsearchterms) | **GET** /asset/search-terms | 
 *AssetApi* | [**getAssetStatistics**](doc//AssetApi.md#getassetstatistics) | **GET** /asset/statistics | 

mobile/openapi/doc/AllJobStatusResponseDto.md

@@ -9,6 +9,7 @@ import 'package:openapi/api.dart';
 Name | Type | Description | Notes
 ------------ | ------------- | ------------- | -------------
 **backgroundTask** | [**JobStatusDto**](JobStatusDto.md) |  | 
+**duplicateDetection** | [**JobStatusDto**](JobStatusDto.md) |  | 
 **faceDetection** | [**JobStatusDto**](JobStatusDto.md) |  | 
 **facialRecognition** | [**JobStatusDto**](JobStatusDto.md) |  | 
 **library_** | [**JobStatusDto**](JobStatusDto.md) |  | 

mobile/openapi/doc/AssetApi.md

@@ -14,6 +14,7 @@ Method | HTTP request | Description
 [**deleteAssets**](AssetApi.md#deleteassets) | **DELETE** /asset | 
 [**getAllAssets**](AssetApi.md#getallassets) | **GET** /asset | 
 [**getAllUserAssetsByDeviceId**](AssetApi.md#getalluserassetsbydeviceid) | **GET** /asset/device/{deviceId} | 
+[**getAssetDuplicates**](AssetApi.md#getassetduplicates) | **GET** /asset/duplicates | 
 [**getAssetInfo**](AssetApi.md#getassetinfo) | **GET** /asset/{id} | 
 [**getAssetSearchTerms**](AssetApi.md#getassetsearchterms) | **GET** /asset/search-terms | 
 [**getAssetStatistics**](AssetApi.md#getassetstatistics) | **GET** /asset/statistics | 
@@ -328,6 +329,57 @@ Name | Type | Description  | Notes
 
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)
 
+# **getAssetDuplicates**
+> List<AssetResponseDto> getAssetDuplicates()
+
+
+
+### Example
+```dart
+import 'package:openapi/api.dart';
+// TODO Configure API key authorization: cookie
+//defaultApiClient.getAuthentication<ApiKeyAuth>('cookie').apiKey = 'YOUR_API_KEY';
+// uncomment below to setup prefix (e.g. Bearer) for API key, if needed
+//defaultApiClient.getAuthentication<ApiKeyAuth>('cookie').apiKeyPrefix = 'Bearer';
+// TODO Configure API key authorization: api_key
+//defaultApiClient.getAuthentication<ApiKeyAuth>('api_key').apiKey = 'YOUR_API_KEY';
+// uncomment below to setup prefix (e.g. Bearer) for API key, if needed
+//defaultApiClient.getAuthentication<ApiKeyAuth>('api_key').apiKeyPrefix = 'Bearer';
+// TODO Configure HTTP Bearer authorization: bearer
+// Case 1. Use String Token
+//defaultApiClient.getAuthentication<HttpBearerAuth>('bearer').setAccessToken('YOUR_ACCESS_TOKEN');
+// Case 2. Use Function which generate token.
+// String yourTokenGeneratorFunction() { ... }
+//defaultApiClient.getAuthentication<HttpBearerAuth>('bearer').setAccessToken(yourTokenGeneratorFunction);
+
+final api_instance = AssetApi();
+
+try {
+    final result = api_instance.getAssetDuplicates();
+    print(result);
+} catch (e) {
+    print('Exception when calling AssetApi->getAssetDuplicates: $e\n');
+}
+```
+
+### Parameters
+This endpoint does not need any parameter.
+
+### Return type
+
+[**List<AssetResponseDto>**](AssetResponseDto.md)
+
+### Authorization
+
+[cookie](../README.md#cookie), [api_key](../README.md#api_key), [bearer](../README.md#bearer)
+
+### HTTP request headers
+
+ - **Content-Type**: Not defined
+ - **Accept**: application/json
+
+[[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)
+
 # **getAssetInfo**
 > AssetResponseDto getAssetInfo(id, key)
 

mobile/openapi/doc/CLIPConfig.md

@@ -8,6 +8,7 @@ import 'package:openapi/api.dart';
 ## Properties
 Name | Type | Description | Notes
 ------------ | ------------- | ------------- | -------------
+**duplicateThreshold** | **double** |  | 
 **enabled** | **bool** |  | 
 **mode** | [**CLIPMode**](CLIPMode.md) |  | [optional] 
 **modelName** | **String** |  | 

mobile/openapi/lib/api/asset_api.dart

@@ -326,6 +326,50 @@ class AssetApi {
     return null;
   }
 
+  /// Performs an HTTP 'GET /asset/duplicates' operation and returns the [Response].
+  Future<Response> getAssetDuplicatesWithHttpInfo() async {
+    // ignore: prefer_const_declarations
+    final path = r'/asset/duplicates';
+
+    // ignore: prefer_final_locals
+    Object? postBody;
+
+    final queryParams = <QueryParam>[];
+    final headerParams = <String, String>{};
+    final formParams = <String, String>{};
+
+    const contentTypes = <String>[];
+
+
+    return apiClient.invokeAPI(
+      path,
+      'GET',
+      queryParams,
+      postBody,
+      headerParams,
+      formParams,
+      contentTypes.isEmpty ? null : contentTypes.first,
+    );
+  }
+
+  Future<List<AssetResponseDto>?> getAssetDuplicates() async {
+    final response = await getAssetDuplicatesWithHttpInfo();
+    if (response.statusCode >= HttpStatus.badRequest) {
+      throw ApiException(response.statusCode, await _decodeBodyBytes(response));
+    }
+    // When a remote server returns no body with a status of 204, we shall not decode it.
+    // At the time of writing this, `dart:convert` will throw an "Unexpected end of input"
+    // FormatException when trying to decode an empty string.
+    if (response.body.isNotEmpty && response.statusCode != HttpStatus.noContent) {
+      final responseBody = await _decodeBodyBytes(response);
+      return (await apiClient.deserializeAsync(responseBody, 'List<AssetResponseDto>') as List)
+        .cast<AssetResponseDto>()
+        .toList(growable: false);
+
+    }
+    return null;
+  }
+
   /// Performs an HTTP 'GET /asset/{id}' operation and returns the [Response].
   /// Parameters:
   ///

mobile/openapi/lib/model/all_job_status_response_dto.dart

@@ -14,6 +14,7 @@ class AllJobStatusResponseDto {
   /// Returns a new [AllJobStatusResponseDto] instance.
   AllJobStatusResponseDto({
     required this.backgroundTask,
+    required this.duplicateDetection,
     required this.faceDetection,
     required this.facialRecognition,
     required this.library_,
@@ -29,6 +30,8 @@ class AllJobStatusResponseDto {
 
   JobStatusDto backgroundTask;
 
+  JobStatusDto duplicateDetection;
+
   JobStatusDto faceDetection;
 
   JobStatusDto facialRecognition;
@@ -54,6 +57,7 @@ class AllJobStatusResponseDto {
   @override
   bool operator ==(Object other) => identical(this, other) || other is AllJobStatusResponseDto &&
     other.backgroundTask == backgroundTask &&
+    other.duplicateDetection == duplicateDetection &&
     other.faceDetection == faceDetection &&
     other.facialRecognition == facialRecognition &&
     other.library_ == library_ &&
@@ -70,6 +74,7 @@ class AllJobStatusResponseDto {
   int get hashCode =>
     // ignore: unnecessary_parenthesis
     (backgroundTask.hashCode) +
+    (duplicateDetection.hashCode) +
     (faceDetection.hashCode) +
     (facialRecognition.hashCode) +
     (library_.hashCode) +
@@ -83,11 +88,12 @@ class AllJobStatusResponseDto {
     (videoConversion.hashCode);
 
   @override
-  String toString() => 'AllJobStatusResponseDto[backgroundTask=$backgroundTask, faceDetection=$faceDetection, facialRecognition=$facialRecognition, library_=$library_, metadataExtraction=$metadataExtraction, migration=$migration, search=$search, sidecar=$sidecar, smartSearch=$smartSearch, storageTemplateMigration=$storageTemplateMigration, thumbnailGeneration=$thumbnailGeneration, videoConversion=$videoConversion]';
+  String toString() => 'AllJobStatusResponseDto[backgroundTask=$backgroundTask, duplicateDetection=$duplicateDetection, faceDetection=$faceDetection, facialRecognition=$facialRecognition, library_=$library_, metadataExtraction=$metadataExtraction, migration=$migration, search=$search, sidecar=$sidecar, smartSearch=$smartSearch, storageTemplateMigration=$storageTemplateMigration, thumbnailGeneration=$thumbnailGeneration, videoConversion=$videoConversion]';
 
   Map<String, dynamic> toJson() {
     final json = <String, dynamic>{};
       json[r'backgroundTask'] = this.backgroundTask;
+      json[r'duplicateDetection'] = this.duplicateDetection;
       json[r'faceDetection'] = this.faceDetection;
       json[r'facialRecognition'] = this.facialRecognition;
       json[r'library'] = this.library_;
@@ -111,6 +117,7 @@ class AllJobStatusResponseDto {
 
       return AllJobStatusResponseDto(
         backgroundTask: JobStatusDto.fromJson(json[r'backgroundTask'])!,
+        duplicateDetection: JobStatusDto.fromJson(json[r'duplicateDetection'])!,
         faceDetection: JobStatusDto.fromJson(json[r'faceDetection'])!,
         facialRecognition: JobStatusDto.fromJson(json[r'facialRecognition'])!,
         library_: JobStatusDto.fromJson(json[r'library'])!,
@@ -170,6 +177,7 @@ class AllJobStatusResponseDto {
   /// The list of required keys that must be present in a JSON.
   static const requiredKeys = <String>{
     'backgroundTask',
+    'duplicateDetection',
     'faceDetection',
     'facialRecognition',
     'library',

mobile/openapi/lib/model/clip_config.dart

@@ -13,12 +13,15 @@ part of openapi.api;
 class CLIPConfig {
   /// Returns a new [CLIPConfig] instance.
   CLIPConfig({
+    required this.duplicateThreshold,
     required this.enabled,
     this.mode,
     required this.modelName,
     this.modelType,
   });
 
+  double duplicateThreshold;
+
   bool enabled;
 
   ///
@@ -41,6 +44,7 @@ class CLIPConfig {
 
   @override
   bool operator ==(Object other) => identical(this, other) || other is CLIPConfig &&
+    other.duplicateThreshold == duplicateThreshold &&
     other.enabled == enabled &&
     other.mode == mode &&
     other.modelName == modelName &&
@@ -49,16 +53,18 @@ class CLIPConfig {
   @override
   int get hashCode =>
     // ignore: unnecessary_parenthesis
+    (duplicateThreshold.hashCode) +
     (enabled.hashCode) +
     (mode == null ? 0 : mode!.hashCode) +
     (modelName.hashCode) +
     (modelType == null ? 0 : modelType!.hashCode);
 
   @override
-  String toString() => 'CLIPConfig[enabled=$enabled, mode=$mode, modelName=$modelName, modelType=$modelType]';
+  String toString() => 'CLIPConfig[duplicateThreshold=$duplicateThreshold, enabled=$enabled, mode=$mode, modelName=$modelName, modelType=$modelType]';
 
   Map<String, dynamic> toJson() {
     final json = <String, dynamic>{};
+      json[r'duplicateThreshold'] = this.duplicateThreshold;
       json[r'enabled'] = this.enabled;
     if (this.mode != null) {
       json[r'mode'] = this.mode;
@@ -82,6 +88,7 @@ class CLIPConfig {
       final json = value.cast<String, dynamic>();
 
       return CLIPConfig(
+        duplicateThreshold: mapValueOfType<double>(json, r'duplicateThreshold')!,
         enabled: mapValueOfType<bool>(json, r'enabled')!,
         mode: CLIPMode.fromJson(json[r'mode']),
         modelName: mapValueOfType<String>(json, r'modelName')!,
@@ -133,6 +140,7 @@ class CLIPConfig {
 
   /// The list of required keys that must be present in a JSON.
   static const requiredKeys = <String>{
+    'duplicateThreshold',
     'enabled',
     'modelName',
   };

mobile/openapi/lib/model/job_name.dart

@@ -29,6 +29,7 @@ class JobName {
   static const faceDetection = JobName._(r'faceDetection');
   static const facialRecognition = JobName._(r'facialRecognition');
   static const smartSearch = JobName._(r'smartSearch');
+  static const duplicateDetection = JobName._(r'duplicateDetection');
   static const backgroundTask = JobName._(r'backgroundTask');
   static const storageTemplateMigration = JobName._(r'storageTemplateMigration');
   static const migration = JobName._(r'migration');
@@ -44,6 +45,7 @@ class JobName {
     faceDetection,
     facialRecognition,
     smartSearch,
+    duplicateDetection,
     backgroundTask,
     storageTemplateMigration,
     migration,
@@ -94,6 +96,7 @@ class JobNameTypeTransformer {
         case r'faceDetection': return JobName.faceDetection;
         case r'facialRecognition': return JobName.facialRecognition;
         case r'smartSearch': return JobName.smartSearch;
+        case r'duplicateDetection': return JobName.duplicateDetection;
         case r'backgroundTask': return JobName.backgroundTask;
         case r'storageTemplateMigration': return JobName.storageTemplateMigration;
         case r'migration': return JobName.migration;

mobile/openapi/test/all_job_status_response_dto_test.dart

@@ -21,6 +21,11 @@ void main() {
       // TODO
     });
 
+    // JobStatusDto duplicateDetection
+    test('to test the property `duplicateDetection`', () async {
+      // TODO
+    });
+
     // JobStatusDto faceDetection
     test('to test the property `faceDetection`', () async {
       // TODO

mobile/openapi/test/asset_api_test.dart

@@ -50,6 +50,11 @@ void main() {
       // TODO
     });
 
+    //Future<List<AssetResponseDto>> getAssetDuplicates() async
+    test('test getAssetDuplicates', () async {
+      // TODO
+    });
+
     //Future<AssetResponseDto> getAssetInfo(String id, { String key }) async
     test('test getAssetInfo', () async {
       // TODO

mobile/openapi/test/clip_config_test.dart

@@ -16,6 +16,11 @@ void main() {
   // final instance = CLIPConfig();
 
   group('test CLIPConfig', () {
+    // double duplicateThreshold
+    test('to test the property `duplicateThreshold`', () async {
+      // TODO
+    });
+
     // bool enabled
     test('to test the property `enabled`', () async {
       // TODO

open-api/immich-openapi-specs.json

@@ -1213,6 +1213,41 @@
         ]
       }
     },
+    "/asset/duplicates": {
+      "get": {
+        "operationId": "getAssetDuplicates",
+        "parameters": [],
+        "responses": {
+          "200": {
+            "content": {
+              "application/json": {
+                "schema": {
+                  "items": {
+                    "$ref": "#/components/schemas/AssetResponseDto"
+                  },
+                  "type": "array"
+                }
+              }
+            },
+            "description": ""
+          }
+        },
+        "security": [
+          {
+            "bearer": []
+          },
+          {
+            "cookie": []
+          },
+          {
+            "api_key": []
+          }
+        ],
+        "tags": [
+          "Asset"
+        ]
+      }
+    },
     "/asset/exist": {
       "post": {
         "description": "Checks if multiple assets exist on the server and returns all existing - used by background backup",
@@ -6831,6 +6866,9 @@
           "backgroundTask": {
             "$ref": "#/components/schemas/JobStatusDto"
           },
+          "duplicateDetection": {
+            "$ref": "#/components/schemas/JobStatusDto"
+          },
           "faceDetection": {
             "$ref": "#/components/schemas/JobStatusDto"
           },
@@ -6867,6 +6905,7 @@
         },
         "required": [
           "backgroundTask",
+          "duplicateDetection",
           "faceDetection",
           "facialRecognition",
           "library",
@@ -7479,6 +7518,10 @@
       },
       "CLIPConfig": {
         "properties": {
+          "duplicateThreshold": {
+            "format": "float",
+            "type": "number"
+          },
           "enabled": {
             "type": "boolean"
           },
@@ -7493,6 +7536,7 @@
           }
         },
         "required": [
+          "duplicateThreshold",
           "enabled",
           "modelName"
         ],
@@ -8178,6 +8222,7 @@
           "faceDetection",
           "facialRecognition",
           "smartSearch",
+          "duplicateDetection",
           "backgroundTask",
           "storageTemplateMigration",
           "migration",

open-api/typescript-sdk/src/fetch-client.ts

@@ -429,6 +429,7 @@ export type JobStatusDto = {
 };
 export type AllJobStatusResponseDto = {
     backgroundTask: JobStatusDto;
+    duplicateDetection: JobStatusDto;
     faceDetection: JobStatusDto;
     facialRecognition: JobStatusDto;
     library: JobStatusDto;
@@ -858,6 +859,7 @@ export type SystemConfigLoggingDto = {
     level: LogLevel;
 };
 export type ClipConfig = {
+    duplicateThreshold: number;
     enabled: boolean;
     mode?: CLIPMode;
     modelName: string;
@@ -1321,6 +1323,14 @@ export function getAllUserAssetsByDeviceId({ deviceId }: {
         ...opts
     }));
 }
+export function getAssetDuplicates(opts?: Oazapfts.RequestOpts) {
+    return oazapfts.ok(oazapfts.fetchJson<{
+        status: 200;
+        data: AssetResponseDto[];
+    }>("/asset/duplicates", {
+        ...opts
+    }));
+}
 /**
  * Checks if multiple assets exist on the server and returns all existing - used by background backup
  */
@@ -2816,6 +2826,7 @@ export enum JobName {
     FaceDetection = "faceDetection",
     FacialRecognition = "facialRecognition",
     SmartSearch = "smartSearch",
+    DuplicateDetection = "duplicateDetection",
     BackgroundTask = "backgroundTask",
     StorageTemplateMigration = "storageTemplateMigration",
     Migration = "migration",

server/src/controllers/asset.controller.ts

@@ -70,6 +70,11 @@ export class AssetController {
     return this.service.getStatistics(auth, dto);
   }
 
+  @Get('duplicates')
+  getAssetDuplicates(@Auth() auth: AuthDto): Promise<AssetResponseDto[]> {
+    return this.service.getDuplicates(auth);
+  }
+
   @Post('jobs')
   @HttpCode(HttpStatus.NO_CONTENT)
   runAssetJobs(@Auth() auth: AuthDto, @Body() dto: AssetJobsDto): Promise<void> {

server/src/cores/system-config.core.ts

@@ -71,6 +71,7 @@ export const defaults = Object.freeze<SystemConfig>({
     clip: {
       enabled: true,
       modelName: 'ViT-B-32__openai',
+      duplicateThreshold: 0.03,
     },
     facialRecognition: {
       enabled: true,

server/src/dtos/job.dto.ts

@@ -73,6 +73,9 @@ export class AllJobStatusResponseDto implements Record<QueueName, JobStatusDto>
   @ApiProperty({ type: JobStatusDto })
   [QueueName.SEARCH]!: JobStatusDto;
 
+  @ApiProperty({ type: JobStatusDto })
+  [QueueName.DUPLICATE_DETECTION]!: JobStatusDto;
+
   @ApiProperty({ type: JobStatusDto })
   [QueueName.FACE_DETECTION]!: JobStatusDto;
 

server/src/dtos/model-config.dto.ts

@@ -23,6 +23,13 @@ export class CLIPConfig extends ModelConfig {
   @Optional()
   @ApiProperty({ enumName: 'CLIPMode', enum: CLIPMode })
   mode?: CLIPMode;
+
+  @IsNumber()
+  @Min(0.01)
+  @Max(0.1)
+  @Type(() => Number)
+  @ApiProperty({ type: 'number', format: 'float' })
+  duplicateThreshold!: number;
 }
 
 export class RecognitionConfig extends ModelConfig {

server/src/entities/asset-duplicate.entity.ts

@@ -0,0 +1,16 @@
+import { AssetEntity } from 'src/entities/asset.entity';
+import { Entity, Index, JoinColumn, OneToMany, PrimaryColumn } from 'typeorm';
+
+@Entity('asset_duplicates')
+@Index('asset_duplicates_assetId_uindex', ['assetId'], { unique: true })
+export class AssetDuplicateEntity {
+  @OneToMany(() => AssetEntity, (asset) => asset.duplicates, { onDelete: 'CASCADE', onUpdate: 'CASCADE' })
+  @JoinColumn({ name: 'assetId', referencedColumnName: 'id' })
+  assets!: AssetEntity;
+
+  @PrimaryColumn()
+  id!: string;
+
+  @PrimaryColumn()
+  assetId!: string;
+}

server/src/entities/asset-job-status.entity.ts

@@ -15,4 +15,7 @@ export class AssetJobStatusEntity {
 
   @Column({ type: 'timestamptz', nullable: true })
   metadataExtractedAt!: Date | null;
+
+  @Column({ type: 'timestamptz', nullable: true })
+  duplicatesDetectedAt!: Date | null;
 }

server/src/entities/asset.entity.ts

@@ -1,4 +1,5 @@
 import { AlbumEntity } from 'src/entities/album.entity';
+import { AssetDuplicateEntity } from 'src/entities/asset-duplicate.entity';
 import { AssetFaceEntity } from 'src/entities/asset-face.entity';
 import { AssetJobStatusEntity } from 'src/entities/asset-job-status.entity';
 import { AssetStackEntity } from 'src/entities/asset-stack.entity';
@@ -168,6 +169,13 @@ export class AssetEntity {
 
   @OneToOne(() => AssetJobStatusEntity, (jobStatus) => jobStatus.asset, { nullable: true })
   jobStatus?: AssetJobStatusEntity;
+
+  @Column({ nullable: true })
+  duplicateId?: string | null;
+
+  @ManyToOne(() => AssetDuplicateEntity, { nullable: true })
+  @JoinColumn({ name: 'duplicateId' })
+  duplicates?: AssetDuplicateEntity | null;
 }
 
 export enum AssetType {

server/src/entities/index.ts

@@ -20,12 +20,14 @@ import { SystemMetadataEntity } from 'src/entities/system-metadata.entity';
 import { TagEntity } from 'src/entities/tag.entity';
 import { UserTokenEntity } from 'src/entities/user-token.entity';
 import { UserEntity } from 'src/entities/user.entity';
+import { AssetDuplicateEntity } from 'src/entities/asset-duplicate.entity';
 
 export const entities = [
   ActivityEntity,
   AlbumEntity,
   APIKeyEntity,
   AssetEntity,
+  AssetDuplicateEntity,
   AssetStackEntity,
   AssetFaceEntity,
   AssetJobStatusEntity,

server/src/entities/smart-search.entity.ts

@@ -11,10 +11,6 @@ export class SmartSearchEntity {
   assetId!: string;
 
   @Index('clip_index', { synchronize: false })
-  @Column({
+  @Column({ type: 'float4', array: true, transformer: { from: (v) => JSON.parse(v), to: (v) => v } })
-    type: 'float4',
-    array: true,
-    select: false,
-  })
   embedding!: number[];
 }

server/src/entities/system-config.entity.ts

@@ -39,7 +39,7 @@ export enum SystemConfigKey {
   JOB_METADATA_EXTRACTION_CONCURRENCY = 'job.metadataExtraction.concurrency',
   JOB_VIDEO_CONVERSION_CONCURRENCY = 'job.videoConversion.concurrency',
   JOB_FACE_DETECTION_CONCURRENCY = 'job.faceDetection.concurrency',
-  JOB_CLIP_ENCODING_CONCURRENCY = 'job.smartSearch.concurrency',
+  JOB_SMART_SEARCH_CONCURRENCY = 'job.smartSearch.concurrency',
   JOB_BACKGROUND_TASK_CONCURRENCY = 'job.backgroundTask.concurrency',
   JOB_STORAGE_TEMPLATE_MIGRATION_CONCURRENCY = 'job.storageTemplateMigration.concurrency',
   JOB_SEARCH_CONCURRENCY = 'job.search.concurrency',
@@ -208,6 +208,7 @@ export interface SystemConfig {
     clip: {
       enabled: boolean;
       modelName: string;
+      duplicateThreshold: number;
     };
     facialRecognition: {
       enabled: boolean;

server/src/interfaces/asset-duplicate.interface.ts

@@ -0,0 +1,9 @@
+import { AssetDuplicateEntity } from 'src/entities/asset-duplicate.entity';
+
+export const IAssetDuplicateRepository = 'IAssetDuplicateRepository';
+
+export interface IAssetDuplicateRepository {
+  upsert(id: string, assetIds: string[], oldDuplicateIds?: string[]): Promise<void>;
+  delete(id: string): Promise<void>;
+  getById(id: string): Promise<AssetDuplicateEntity | null>;
+}

server/src/interfaces/asset.interface.ts

@@ -40,6 +40,7 @@ export enum WithoutProperty {
   ENCODED_VIDEO = 'encoded-video',
   EXIF = 'exif',
   SMART_SEARCH = 'smart-search',
+  DUPLICATE = 'duplicate',
   OBJECT_TAGS = 'object-tags',
   FACES = 'faces',
   PERSON = 'person',
@@ -60,6 +61,7 @@ export interface AssetBuilderOptions {
   isArchived?: boolean;
   isFavorite?: boolean;
   isTrashed?: boolean;
+  isDuplicate?: boolean;
   albumId?: string;
   personId?: string;
   userIds?: string[];
@@ -171,8 +173,9 @@ export interface IAssetRepository {
   getTimeBuckets(options: TimeBucketOptions): Promise<TimeBucketItem[]>;
   getTimeBucket(timeBucket: string, options: TimeBucketOptions): Promise<AssetEntity[]>;
   upsertExif(exif: Partial<ExifEntity>): Promise<void>;
-  upsertJobStatus(jobStatus: Partial<AssetJobStatusEntity>): Promise<void>;
+  upsertJobStatus(jobStatus: Partial<AssetJobStatusEntity> | Partial<AssetJobStatusEntity>[]): Promise<void>;
   getAssetIdByCity(userId: string, options: AssetExploreFieldOptions): Promise<SearchExploreItem<string>>;
   getAssetIdByTag(userId: string, options: AssetExploreFieldOptions): Promise<SearchExploreItem<string>>;
+  getDuplicates(options: AssetBuilderOptions): Promise<AssetEntity[]>;
   searchMetadata(query: string, userIds: string[], options: MetadataSearchOptions): Promise<AssetEntity[]>;
 }

server/src/interfaces/job.interface.ts

@@ -5,6 +5,7 @@ export enum QueueName {
   FACE_DETECTION = 'faceDetection',
   FACIAL_RECOGNITION = 'facialRecognition',
   SMART_SEARCH = 'smartSearch',
+  DUPLICATE_DETECTION = 'duplicateDetection',
   BACKGROUND_TASK = 'backgroundTask',
   STORAGE_TEMPLATE_MIGRATION = 'storageTemplateMigration',
   MIGRATION = 'migration',
@@ -15,7 +16,7 @@ export enum QueueName {
 
 export type ConcurrentQueueName = Exclude<
   QueueName,
-  QueueName.STORAGE_TEMPLATE_MIGRATION | QueueName.FACIAL_RECOGNITION
+  QueueName.STORAGE_TEMPLATE_MIGRATION | QueueName.FACIAL_RECOGNITION | QueueName.DUPLICATE_DETECTION
 >;
 
 export enum JobCommand {
@@ -84,6 +85,10 @@ export enum JobName {
   QUEUE_SMART_SEARCH = 'queue-smart-search',
   SMART_SEARCH = 'smart-search',
 
+  // duplicate detection
+  QUEUE_DUPLICATE_DETECTION = 'queue-duplicate-detection',
+  DUPLICATE_DETECTION = 'duplicate-detection',
+
   // XMP sidecars
   QUEUE_SIDECAR = 'queue-sidecar',
   SIDECAR_DISCOVERY = 'sidecar-discovery',
@@ -199,6 +204,10 @@ export type JobItem =
   | { name: JobName.QUEUE_SMART_SEARCH; data: IBaseJob }
   | { name: JobName.SMART_SEARCH; data: IEntityJob }
 
+  // Duplicate Detection
+  | { name: JobName.QUEUE_DUPLICATE_DETECTION; data: IBaseJob }
+  | { name: JobName.DUPLICATE_DETECTION; data: IEntityJob }
+
   // Filesystem
   | { name: JobName.DELETE_FILES; data: IDeleteFilesJob }
 

server/src/interfaces/search.interface.ts

@@ -177,15 +177,29 @@ export interface FaceEmbeddingSearch extends SearchEmbeddingOptions {
   maxDistance?: number;
 }
 
+export interface AssetDuplicateSearch {
+  assetId: string;
+  embedding: Embedding;
+  userIds: string[];
+  maxDistance?: number;
+}
+
 export interface FaceSearchResult {
   distance: number;
   face: AssetFaceEntity;
 }
 
+export interface AssetDuplicateResult {
+  assetId: string;
+  duplicateId: string;
+  distance: number;
+}
+
 export interface ISearchRepository {
   init(modelName: string): Promise<void>;
   searchMetadata(pagination: SearchPaginationOptions, options: AssetSearchOptions): Paginated<AssetEntity>;
   searchSmart(pagination: SearchPaginationOptions, options: SmartSearchOptions): Paginated<AssetEntity>;
+  searchDuplicates(options: AssetDuplicateSearch): Promise<AssetDuplicateResult[]>;
   searchFaces(search: FaceEmbeddingSearch): Promise<FaceSearchResult[]>;
   upsert(assetId: string, embedding: number[]): Promise<void>;
   searchPlaces(placeName: string): Promise<GeodataPlacesEntity[]>;

server/src/migrations/1711989989911-CreateAssetDuplicateTable.ts

@@ -0,0 +1,36 @@
+import { MigrationInterface, QueryRunner } from 'typeorm';
+
+export class CreateAssetDuplicateTable1711989989911 implements MigrationInterface {
+  public async up(queryRunner: QueryRunner): Promise<void> {
+    await queryRunner.query(`
+        CREATE TABLE asset_duplicates (
+            id uuid,
+            "assetId" uuid REFERENCES assets ON UPDATE CASCADE ON DELETE CASCADE,
+            PRIMARY KEY (id, "assetId")
+        );
+    `);
+
+    await queryRunner.query(`ALTER TABLE assets ADD COLUMN "duplicateId" uuid`);
+
+    await queryRunner.query(`ALTER TABLE asset_job_status ADD COLUMN "duplicatesDetectedAt" timestamptz`);
+
+    await queryRunner.query(`
+        ALTER TABLE assets
+          ADD CONSTRAINT asset_duplicates_id
+            FOREIGN KEY ("duplicateId", id)
+            REFERENCES asset_duplicates DEFERRABLE INITIALLY DEFERRED
+    `);
+
+    await queryRunner.query(`
+        CREATE UNIQUE INDEX "asset_duplicates_assetId_uindex"
+        ON asset_duplicates ("assetId")
+    `);
+
+    await queryRunner.query(`CREATE INDEX "IDX_assets_duplicateId" ON assets ("duplicateId")`);
+  }
+
+  public async down(queryRunner: QueryRunner): Promise<void> {
+    await queryRunner.query(`ALTER TABLE assets DROP COLUMN "duplicateId"`);
+    await queryRunner.query(`DROP TABLE asset_duplicates`);
+  }
+}

server/src/queries/asset.repository.sql

@@ -31,6 +31,7 @@ SELECT
   "entity"."originalFileName" AS "entity_originalFileName",
   "entity"."sidecarPath" AS "entity_sidecarPath",
   "entity"."stackId" AS "entity_stackId",
+  "entity"."duplicateId" AS "entity_duplicateId",
   "exifInfo"."assetId" AS "exifInfo_assetId",
   "exifInfo"."description" AS "exifInfo_description",
   "exifInfo"."exifImageWidth" AS "exifInfo_exifImageWidth",
@@ -113,7 +114,8 @@ SELECT
   "AssetEntity"."livePhotoVideoId" AS "AssetEntity_livePhotoVideoId",
   "AssetEntity"."originalFileName" AS "AssetEntity_originalFileName",
   "AssetEntity"."sidecarPath" AS "AssetEntity_sidecarPath",
-  "AssetEntity"."stackId" AS "AssetEntity_stackId"
+  "AssetEntity"."stackId" AS "AssetEntity_stackId",
+  "AssetEntity"."duplicateId" AS "AssetEntity_duplicateId"
 FROM
   "assets" "AssetEntity"
 WHERE
@@ -150,6 +152,7 @@ SELECT
   "AssetEntity"."originalFileName" AS "AssetEntity_originalFileName",
   "AssetEntity"."sidecarPath" AS "AssetEntity_sidecarPath",
   "AssetEntity"."stackId" AS "AssetEntity_stackId",
+  "AssetEntity"."duplicateId" AS "AssetEntity_duplicateId",
   "AssetEntity__AssetEntity_exifInfo"."assetId" AS "AssetEntity__AssetEntity_exifInfo_assetId",
   "AssetEntity__AssetEntity_exifInfo"."description" AS "AssetEntity__AssetEntity_exifInfo_description",
   "AssetEntity__AssetEntity_exifInfo"."exifImageWidth" AS "AssetEntity__AssetEntity_exifInfo_exifImageWidth",
@@ -234,7 +237,8 @@ SELECT
   "bd93d5747511a4dad4923546c51365bf1a803774"."livePhotoVideoId" AS "bd93d5747511a4dad4923546c51365bf1a803774_livePhotoVideoId",
   "bd93d5747511a4dad4923546c51365bf1a803774"."originalFileName" AS "bd93d5747511a4dad4923546c51365bf1a803774_originalFileName",
   "bd93d5747511a4dad4923546c51365bf1a803774"."sidecarPath" AS "bd93d5747511a4dad4923546c51365bf1a803774_sidecarPath",
-  "bd93d5747511a4dad4923546c51365bf1a803774"."stackId" AS "bd93d5747511a4dad4923546c51365bf1a803774_stackId"
+  "bd93d5747511a4dad4923546c51365bf1a803774"."stackId" AS "bd93d5747511a4dad4923546c51365bf1a803774_stackId",
+  "bd93d5747511a4dad4923546c51365bf1a803774"."duplicateId" AS "bd93d5747511a4dad4923546c51365bf1a803774_duplicateId"
 FROM
   "assets" "AssetEntity"
   LEFT JOIN "exif" "AssetEntity__AssetEntity_exifInfo" ON "AssetEntity__AssetEntity_exifInfo"."assetId" = "AssetEntity"."id"
@@ -315,7 +319,8 @@ FROM
       "AssetEntity"."livePhotoVideoId" AS "AssetEntity_livePhotoVideoId",
       "AssetEntity"."originalFileName" AS "AssetEntity_originalFileName",
       "AssetEntity"."sidecarPath" AS "AssetEntity_sidecarPath",
-      "AssetEntity"."stackId" AS "AssetEntity_stackId"
+      "AssetEntity"."stackId" AS "AssetEntity_stackId",
+      "AssetEntity"."duplicateId" AS "AssetEntity_duplicateId"
     FROM
       "assets" "AssetEntity"
       LEFT JOIN "libraries" "AssetEntity__AssetEntity_library" ON "AssetEntity__AssetEntity_library"."id" = "AssetEntity"."libraryId"
@@ -412,7 +417,8 @@ SELECT
   "AssetEntity"."livePhotoVideoId" AS "AssetEntity_livePhotoVideoId",
   "AssetEntity"."originalFileName" AS "AssetEntity_originalFileName",
   "AssetEntity"."sidecarPath" AS "AssetEntity_sidecarPath",
-  "AssetEntity"."stackId" AS "AssetEntity_stackId"
+  "AssetEntity"."stackId" AS "AssetEntity_stackId",
+  "AssetEntity"."duplicateId" AS "AssetEntity_duplicateId"
 FROM
   "assets" "AssetEntity"
 WHERE
@@ -458,7 +464,8 @@ SELECT
   "AssetEntity"."livePhotoVideoId" AS "AssetEntity_livePhotoVideoId",
   "AssetEntity"."originalFileName" AS "AssetEntity_originalFileName",
   "AssetEntity"."sidecarPath" AS "AssetEntity_sidecarPath",
-  "AssetEntity"."stackId" AS "AssetEntity_stackId"
+  "AssetEntity"."stackId" AS "AssetEntity_stackId",
+  "AssetEntity"."duplicateId" AS "AssetEntity_duplicateId"
 FROM
   "assets" "AssetEntity"
 WHERE
@@ -502,7 +509,8 @@ SELECT
   "AssetEntity"."livePhotoVideoId" AS "AssetEntity_livePhotoVideoId",
   "AssetEntity"."originalFileName" AS "AssetEntity_originalFileName",
   "AssetEntity"."sidecarPath" AS "AssetEntity_sidecarPath",
-  "AssetEntity"."stackId" AS "AssetEntity_stackId"
+  "AssetEntity"."stackId" AS "AssetEntity_stackId",
+  "AssetEntity"."duplicateId" AS "AssetEntity_duplicateId"
 FROM
   "assets" "AssetEntity"
 WHERE
@@ -592,6 +600,7 @@ SELECT
   "asset"."originalFileName" AS "asset_originalFileName",
   "asset"."sidecarPath" AS "asset_sidecarPath",
   "asset"."stackId" AS "asset_stackId",
+  "asset"."duplicateId" AS "asset_duplicateId",
   "exifInfo"."assetId" AS "exifInfo_assetId",
   "exifInfo"."description" AS "exifInfo_description",
   "exifInfo"."exifImageWidth" AS "exifInfo_exifImageWidth",
@@ -650,7 +659,8 @@ SELECT
   "stackedAssets"."livePhotoVideoId" AS "stackedAssets_livePhotoVideoId",
   "stackedAssets"."originalFileName" AS "stackedAssets_originalFileName",
   "stackedAssets"."sidecarPath" AS "stackedAssets_sidecarPath",
-  "stackedAssets"."stackId" AS "stackedAssets_stackId"
+  "stackedAssets"."stackId" AS "stackedAssets_stackId",
+  "stackedAssets"."duplicateId" AS "stackedAssets_duplicateId"
 FROM
   "assets" "asset"
   LEFT JOIN "exif" "exifInfo" ON "exifInfo"."assetId" = "asset"."id"
@@ -677,6 +687,114 @@ ORDER BY
   )::timestamptz DESC,
   "asset"."fileCreatedAt" DESC
 
+-- AssetRepository.getDuplicates
+SELECT
+  "asset"."id" AS "asset_id",
+  "asset"."deviceAssetId" AS "asset_deviceAssetId",
+  "asset"."ownerId" AS "asset_ownerId",
+  "asset"."libraryId" AS "asset_libraryId",
+  "asset"."deviceId" AS "asset_deviceId",
+  "asset"."type" AS "asset_type",
+  "asset"."originalPath" AS "asset_originalPath",
+  "asset"."resizePath" AS "asset_resizePath",
+  "asset"."webpPath" AS "asset_webpPath",
+  "asset"."thumbhash" AS "asset_thumbhash",
+  "asset"."encodedVideoPath" AS "asset_encodedVideoPath",
+  "asset"."createdAt" AS "asset_createdAt",
+  "asset"."updatedAt" AS "asset_updatedAt",
+  "asset"."deletedAt" AS "asset_deletedAt",
+  "asset"."fileCreatedAt" AS "asset_fileCreatedAt",
+  "asset"."localDateTime" AS "asset_localDateTime",
+  "asset"."fileModifiedAt" AS "asset_fileModifiedAt",
+  "asset"."isFavorite" AS "asset_isFavorite",
+  "asset"."isArchived" AS "asset_isArchived",
+  "asset"."isExternal" AS "asset_isExternal",
+  "asset"."isReadOnly" AS "asset_isReadOnly",
+  "asset"."isOffline" AS "asset_isOffline",
+  "asset"."checksum" AS "asset_checksum",
+  "asset"."duration" AS "asset_duration",
+  "asset"."isVisible" AS "asset_isVisible",
+  "asset"."livePhotoVideoId" AS "asset_livePhotoVideoId",
+  "asset"."originalFileName" AS "asset_originalFileName",
+  "asset"."sidecarPath" AS "asset_sidecarPath",
+  "asset"."stackId" AS "asset_stackId",
+  "asset"."duplicateId" AS "asset_duplicateId",
+  "exifInfo"."assetId" AS "exifInfo_assetId",
+  "exifInfo"."description" AS "exifInfo_description",
+  "exifInfo"."exifImageWidth" AS "exifInfo_exifImageWidth",
+  "exifInfo"."exifImageHeight" AS "exifInfo_exifImageHeight",
+  "exifInfo"."fileSizeInByte" AS "exifInfo_fileSizeInByte",
+  "exifInfo"."orientation" AS "exifInfo_orientation",
+  "exifInfo"."dateTimeOriginal" AS "exifInfo_dateTimeOriginal",
+  "exifInfo"."modifyDate" AS "exifInfo_modifyDate",
+  "exifInfo"."timeZone" AS "exifInfo_timeZone",
+  "exifInfo"."latitude" AS "exifInfo_latitude",
+  "exifInfo"."longitude" AS "exifInfo_longitude",
+  "exifInfo"."projectionType" AS "exifInfo_projectionType",
+  "exifInfo"."city" AS "exifInfo_city",
+  "exifInfo"."livePhotoCID" AS "exifInfo_livePhotoCID",
+  "exifInfo"."autoStackId" AS "exifInfo_autoStackId",
+  "exifInfo"."state" AS "exifInfo_state",
+  "exifInfo"."country" AS "exifInfo_country",
+  "exifInfo"."make" AS "exifInfo_make",
+  "exifInfo"."model" AS "exifInfo_model",
+  "exifInfo"."lensModel" AS "exifInfo_lensModel",
+  "exifInfo"."fNumber" AS "exifInfo_fNumber",
+  "exifInfo"."focalLength" AS "exifInfo_focalLength",
+  "exifInfo"."iso" AS "exifInfo_iso",
+  "exifInfo"."exposureTime" AS "exifInfo_exposureTime",
+  "exifInfo"."profileDescription" AS "exifInfo_profileDescription",
+  "exifInfo"."colorspace" AS "exifInfo_colorspace",
+  "exifInfo"."bitsPerSample" AS "exifInfo_bitsPerSample",
+  "exifInfo"."fps" AS "exifInfo_fps",
+  "stack"."id" AS "stack_id",
+  "stack"."primaryAssetId" AS "stack_primaryAssetId",
+  "stackedAssets"."id" AS "stackedAssets_id",
+  "stackedAssets"."deviceAssetId" AS "stackedAssets_deviceAssetId",
+  "stackedAssets"."ownerId" AS "stackedAssets_ownerId",
+  "stackedAssets"."libraryId" AS "stackedAssets_libraryId",
+  "stackedAssets"."deviceId" AS "stackedAssets_deviceId",
+  "stackedAssets"."type" AS "stackedAssets_type",
+  "stackedAssets"."originalPath" AS "stackedAssets_originalPath",
+  "stackedAssets"."resizePath" AS "stackedAssets_resizePath",
+  "stackedAssets"."webpPath" AS "stackedAssets_webpPath",
+  "stackedAssets"."thumbhash" AS "stackedAssets_thumbhash",
+  "stackedAssets"."encodedVideoPath" AS "stackedAssets_encodedVideoPath",
+  "stackedAssets"."createdAt" AS "stackedAssets_createdAt",
+  "stackedAssets"."updatedAt" AS "stackedAssets_updatedAt",
+  "stackedAssets"."deletedAt" AS "stackedAssets_deletedAt",
+  "stackedAssets"."fileCreatedAt" AS "stackedAssets_fileCreatedAt",
+  "stackedAssets"."localDateTime" AS "stackedAssets_localDateTime",
+  "stackedAssets"."fileModifiedAt" AS "stackedAssets_fileModifiedAt",
+  "stackedAssets"."isFavorite" AS "stackedAssets_isFavorite",
+  "stackedAssets"."isArchived" AS "stackedAssets_isArchived",
+  "stackedAssets"."isExternal" AS "stackedAssets_isExternal",
+  "stackedAssets"."isReadOnly" AS "stackedAssets_isReadOnly",
+  "stackedAssets"."isOffline" AS "stackedAssets_isOffline",
+  "stackedAssets"."checksum" AS "stackedAssets_checksum",
+  "stackedAssets"."duration" AS "stackedAssets_duration",
+  "stackedAssets"."isVisible" AS "stackedAssets_isVisible",
+  "stackedAssets"."livePhotoVideoId" AS "stackedAssets_livePhotoVideoId",
+  "stackedAssets"."originalFileName" AS "stackedAssets_originalFileName",
+  "stackedAssets"."sidecarPath" AS "stackedAssets_sidecarPath",
+  "stackedAssets"."stackId" AS "stackedAssets_stackId",
+  "stackedAssets"."duplicateId" AS "stackedAssets_duplicateId"
+FROM
+  "assets" "asset"
+  LEFT JOIN "exif" "exifInfo" ON "exifInfo"."assetId" = "asset"."id"
+  LEFT JOIN "asset_stack" "stack" ON "stack"."id" = "asset"."stackId"
+  LEFT JOIN "assets" "stackedAssets" ON "stackedAssets"."stackId" = "stack"."id"
+  AND ("stackedAssets"."deletedAt" IS NULL)
+WHERE
+  (
+    "asset"."isVisible" = true
+    AND "asset"."ownerId" IN ($1, $2)
+    AND "asset"."duplicateId" IS NOT NULL
+  )
+  AND ("asset"."deletedAt" IS NULL)
+ORDER BY
+  "asset"."duplicateId" ASC
+
 -- AssetRepository.getAssetIdByCity
 WITH
   "cities" AS (

server/src/queries/person.repository.sql

@@ -173,7 +173,8 @@ FROM
       "AssetFaceEntity__AssetFaceEntity_asset"."livePhotoVideoId" AS "AssetFaceEntity__AssetFaceEntity_asset_livePhotoVideoId",
       "AssetFaceEntity__AssetFaceEntity_asset"."originalFileName" AS "AssetFaceEntity__AssetFaceEntity_asset_originalFileName",
       "AssetFaceEntity__AssetFaceEntity_asset"."sidecarPath" AS "AssetFaceEntity__AssetFaceEntity_asset_sidecarPath",
-      "AssetFaceEntity__AssetFaceEntity_asset"."stackId" AS "AssetFaceEntity__AssetFaceEntity_asset_stackId"
+      "AssetFaceEntity__AssetFaceEntity_asset"."stackId" AS "AssetFaceEntity__AssetFaceEntity_asset_stackId",
+      "AssetFaceEntity__AssetFaceEntity_asset"."duplicateId" AS "AssetFaceEntity__AssetFaceEntity_asset_duplicateId"
     FROM
       "asset_faces" "AssetFaceEntity"
       LEFT JOIN "person" "AssetFaceEntity__AssetFaceEntity_person" ON "AssetFaceEntity__AssetFaceEntity_person"."id" = "AssetFaceEntity"."personId"
@@ -272,6 +273,7 @@ FROM
       "AssetEntity"."originalFileName" AS "AssetEntity_originalFileName",
       "AssetEntity"."sidecarPath" AS "AssetEntity_sidecarPath",
       "AssetEntity"."stackId" AS "AssetEntity_stackId",
+      "AssetEntity"."duplicateId" AS "AssetEntity_duplicateId",
       "AssetEntity__AssetEntity_faces"."id" AS "AssetEntity__AssetEntity_faces_id",
       "AssetEntity__AssetEntity_faces"."assetId" AS "AssetEntity__AssetEntity_faces_assetId",
       "AssetEntity__AssetEntity_faces"."personId" AS "AssetEntity__AssetEntity_faces_personId",
@@ -401,7 +403,8 @@ SELECT
   "AssetFaceEntity__AssetFaceEntity_asset"."livePhotoVideoId" AS "AssetFaceEntity__AssetFaceEntity_asset_livePhotoVideoId",
   "AssetFaceEntity__AssetFaceEntity_asset"."originalFileName" AS "AssetFaceEntity__AssetFaceEntity_asset_originalFileName",
   "AssetFaceEntity__AssetFaceEntity_asset"."sidecarPath" AS "AssetFaceEntity__AssetFaceEntity_asset_sidecarPath",
-  "AssetFaceEntity__AssetFaceEntity_asset"."stackId" AS "AssetFaceEntity__AssetFaceEntity_asset_stackId"
+  "AssetFaceEntity__AssetFaceEntity_asset"."stackId" AS "AssetFaceEntity__AssetFaceEntity_asset_stackId",
+  "AssetFaceEntity__AssetFaceEntity_asset"."duplicateId" AS "AssetFaceEntity__AssetFaceEntity_asset_duplicateId"
 FROM
   "asset_faces" "AssetFaceEntity"
   LEFT JOIN "assets" "AssetFaceEntity__AssetFaceEntity_asset" ON "AssetFaceEntity__AssetFaceEntity_asset"."id" = "AssetFaceEntity"."assetId"

server/src/queries/search.repository.sql

@@ -36,6 +36,7 @@ FROM
       "asset"."originalFileName" AS "asset_originalFileName",
       "asset"."sidecarPath" AS "asset_sidecarPath",
       "asset"."stackId" AS "asset_stackId",
+      "asset"."duplicateId" AS "asset_duplicateId",
       "stack"."id" AS "stack_id",
       "stack"."primaryAssetId" AS "stack_primaryAssetId",
       "stackedAssets"."id" AS "stackedAssets_id",
@@ -66,7 +67,8 @@ FROM
       "stackedAssets"."livePhotoVideoId" AS "stackedAssets_livePhotoVideoId",
       "stackedAssets"."originalFileName" AS "stackedAssets_originalFileName",
       "stackedAssets"."sidecarPath" AS "stackedAssets_sidecarPath",
-      "stackedAssets"."stackId" AS "stackedAssets_stackId"
+      "stackedAssets"."stackId" AS "stackedAssets_stackId",
+      "stackedAssets"."duplicateId" AS "stackedAssets_duplicateId"
     FROM
       "assets" "asset"
       LEFT JOIN "exif" "exifInfo" ON "exifInfo"."assetId" = "asset"."id"
@@ -132,6 +134,7 @@ SELECT
   "asset"."originalFileName" AS "asset_originalFileName",
   "asset"."sidecarPath" AS "asset_sidecarPath",
   "asset"."stackId" AS "asset_stackId",
+  "asset"."duplicateId" AS "asset_duplicateId",
   "stack"."id" AS "stack_id",
   "stack"."primaryAssetId" AS "stack_primaryAssetId",
   "stackedAssets"."id" AS "stackedAssets_id",
@@ -162,7 +165,8 @@ SELECT
   "stackedAssets"."livePhotoVideoId" AS "stackedAssets_livePhotoVideoId",
   "stackedAssets"."originalFileName" AS "stackedAssets_originalFileName",
   "stackedAssets"."sidecarPath" AS "stackedAssets_sidecarPath",
-  "stackedAssets"."stackId" AS "stackedAssets_stackId"
+  "stackedAssets"."stackId" AS "stackedAssets_stackId",
+  "stackedAssets"."duplicateId" AS "stackedAssets_duplicateId"
 FROM
   "assets" "asset"
   LEFT JOIN "exif" "exifInfo" ON "exifInfo"."assetId" = "asset"."id"
@@ -189,6 +193,49 @@ LIMIT
   101
 COMMIT
 
+-- SearchRepository.searchDuplicates
+WITH
+  "cte" AS (
+    SELECT
+      "asset"."duplicateId" AS "duplicateId",
+      "search"."assetId" AS "assetId",
+      (
+        SELECT
+          embedding
+        FROM
+          smart_search
+        WHERE
+          "assetId" = $1
+      ) <= > "search"."embedding" AS "distance"
+    FROM
+      "assets" "asset"
+      INNER JOIN "smart_search" "search" ON "search"."assetId" = "asset"."id"
+    WHERE
+      (
+        "asset"."ownerId" IN ($2)
+        AND "asset"."id" != $1
+        AND "asset"."isVisible" = $3
+      )
+      AND ("asset"."deletedAt" IS NULL)
+    ORDER BY
+      "search"."embedding" <= > (
+        SELECT
+          embedding
+        FROM
+          smart_search
+        WHERE
+          "assetId" = $1
+      ) ASC
+    LIMIT
+      64
+  )
+SELECT
+  res.*
+FROM
+  "cte" "res"
+WHERE
+  res.distance <= $4
+
 -- SearchRepository.searchFaces
 START TRANSACTION
 SET
@@ -342,6 +389,7 @@ SELECT
   "asset"."originalFileName" AS "asset_originalFileName",
   "asset"."sidecarPath" AS "asset_sidecarPath",
   "asset"."stackId" AS "asset_stackId",
+  "asset"."duplicateId" AS "asset_duplicateId",
   "exif"."assetId" AS "exif_assetId",
   "exif"."description" AS "exif_description",
   "exif"."exifImageWidth" AS "exif_exifImageWidth",

server/src/queries/shared.link.repository.sql

@@ -50,6 +50,7 @@ FROM
       "SharedLinkEntity__SharedLinkEntity_assets"."originalFileName" AS "SharedLinkEntity__SharedLinkEntity_assets_originalFileName",
       "SharedLinkEntity__SharedLinkEntity_assets"."sidecarPath" AS "SharedLinkEntity__SharedLinkEntity_assets_sidecarPath",
       "SharedLinkEntity__SharedLinkEntity_assets"."stackId" AS "SharedLinkEntity__SharedLinkEntity_assets_stackId",
+      "SharedLinkEntity__SharedLinkEntity_assets"."duplicateId" AS "SharedLinkEntity__SharedLinkEntity_assets_duplicateId",
       "9b1d35b344d838023994a3233afd6ffe098be6d8"."assetId" AS "9b1d35b344d838023994a3233afd6ffe098be6d8_assetId",
       "9b1d35b344d838023994a3233afd6ffe098be6d8"."description" AS "9b1d35b344d838023994a3233afd6ffe098be6d8_description",
       "9b1d35b344d838023994a3233afd6ffe098be6d8"."exifImageWidth" AS "9b1d35b344d838023994a3233afd6ffe098be6d8_exifImageWidth",
@@ -117,6 +118,7 @@ FROM
       "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6"."originalFileName" AS "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6_originalFileName",
       "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6"."sidecarPath" AS "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6_sidecarPath",
       "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6"."stackId" AS "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6_stackId",
+      "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6"."duplicateId" AS "4a35f463ae8c5544ede95c4b6d9ce8c686b6bfe6_duplicateId",
       "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f"."assetId" AS "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f_assetId",
       "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f"."description" AS "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f_description",
       "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f"."exifImageWidth" AS "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f_exifImageWidth",
@@ -240,6 +242,7 @@ SELECT
   "SharedLinkEntity__SharedLinkEntity_assets"."originalFileName" AS "SharedLinkEntity__SharedLinkEntity_assets_originalFileName",
   "SharedLinkEntity__SharedLinkEntity_assets"."sidecarPath" AS "SharedLinkEntity__SharedLinkEntity_assets_sidecarPath",
   "SharedLinkEntity__SharedLinkEntity_assets"."stackId" AS "SharedLinkEntity__SharedLinkEntity_assets_stackId",
+  "SharedLinkEntity__SharedLinkEntity_assets"."duplicateId" AS "SharedLinkEntity__SharedLinkEntity_assets_duplicateId",
   "SharedLinkEntity__SharedLinkEntity_album"."id" AS "SharedLinkEntity__SharedLinkEntity_album_id",
   "SharedLinkEntity__SharedLinkEntity_album"."ownerId" AS "SharedLinkEntity__SharedLinkEntity_album_ownerId",
   "SharedLinkEntity__SharedLinkEntity_album"."albumName" AS "SharedLinkEntity__SharedLinkEntity_album_albumName",

server/src/repositories/asset-duplicate.repository.ts

@@ -0,0 +1,43 @@
+import { Injectable } from '@nestjs/common';
+import { InjectRepository } from '@nestjs/typeorm';
+import { AssetDuplicateEntity } from 'src/entities/asset-duplicate.entity';
+import { AssetEntity } from 'src/entities/asset.entity';
+import { IAssetDuplicateRepository } from 'src/interfaces/asset-duplicate.interface';
+import { Instrumentation } from 'src/utils/instrumentation';
+import { In, Repository } from 'typeorm';
+
+@Instrumentation()
+@Injectable()
+export class AssetDuplicateRepository implements IAssetDuplicateRepository {
+  constructor(@InjectRepository(AssetDuplicateEntity) private repository: Repository<AssetDuplicateEntity>) {}
+
+  async upsert(id: string, assetIds: string[], oldDuplicateIds: string[] = []): Promise<void> {
+    await this.repository.manager.transaction(async (manager) => {
+      await manager.upsert(
+        AssetDuplicateEntity,
+        assetIds.map((assetId) => ({ id, assetId })),
+        ['assetId'],
+      );
+      if (oldDuplicateIds.length > 0) {
+        await manager.update(AssetDuplicateEntity, { id: In(oldDuplicateIds) }, { id });
+      }
+      await manager.update(AssetEntity, { id: In(assetIds) }, { duplicateId: id });
+      await manager.update(AssetEntity, { duplicateId: In(oldDuplicateIds) }, { duplicateId: id });
+    });
+  }
+
+  async delete(id: string): Promise<void> {
+    await this.repository.delete(id);
+  }
+
+  async getById(id: string): Promise<AssetDuplicateEntity | null> {
+    return this.repository.findOne({
+      where: {
+        id,
+      },
+      relations: {
+        assets: true,
+      },
+    });
+  }
+}

server/src/repositories/asset.repository.ts

@@ -67,7 +67,7 @@ export class AssetRepository implements IAssetRepository {
     await this.exifRepository.upsert(exif, { conflictPaths: ['assetId'] });
   }
 
-  async upsertJobStatus(jobStatus: Partial<AssetJobStatusEntity>): Promise<void> {
+  async upsertJobStatus(jobStatus: Partial<AssetJobStatusEntity> | Partial<AssetJobStatusEntity>[]): Promise<void> {
     await this.jobStatusRepository.upsert(jobStatus, { conflictPaths: ['assetId'] });
   }
 
@@ -201,6 +201,7 @@ export class AssetRepository implements IAssetRepository {
     let builder = this.repository.createQueryBuilder('asset');
     builder = searchAssetBuilder(builder, options);
     builder.orderBy('asset.createdAt', options.orderDirection ?? 'ASC');
+    builder.innerJoin('asset.smartSearch', 'smartSearch');
     return paginatedBuilder<AssetEntity>(builder, {
       mode: PaginationMode.SKIP_TAKE,
       skip: pagination.skip,
@@ -347,6 +348,18 @@ export class AssetRepository implements IAssetRepository {
         break;
       }
 
+      case WithoutProperty.DUPLICATE: {
+        where = {
+          resizePath: Not(IsNull()),
+          isVisible: true,
+          smartSearch: true,
+          jobStatus: {
+            duplicatesDetectedAt: IsNull(),
+          },
+        };
+        break;
+      }
+
       case WithoutProperty.OBJECT_TAGS: {
         relations = {
           smartInfo: true,
@@ -579,6 +592,13 @@ export class AssetRepository implements IAssetRepository {
     );
   }
 
+  @GenerateSql({ params: [{ userIds: [DummyValue.UUID, DummyValue.UUID] }] })
+  getDuplicates(options: AssetBuilderOptions): Promise<AssetEntity[]> {
+    return this.getBuilder({ ...options, isDuplicate: true })
+      .orderBy('asset.duplicateId')
+      .getMany();
+  }
+
   @GenerateSql({ params: [DummyValue.UUID, { minAssetsPerField: 5, maxFields: 12 }] })
   async getAssetIdByCity(
     ownerId: string,
@@ -638,16 +658,14 @@ export class AssetRepository implements IAssetRepository {
   }
 
   private getBuilder(options: AssetBuilderOptions) {
-    const { isArchived, isFavorite, isTrashed, albumId, personId, userIds, withStacked, exifInfo, assetType } = options;
-
     let builder = this.repository.createQueryBuilder('asset').where('asset.isVisible = true');
-    if (assetType !== undefined) {
+    if (options.assetType !== undefined) {
-      builder = builder.andWhere('asset.type = :assetType', { assetType });
+      builder = builder.andWhere('asset.type = :assetType', { assetType: options.assetType });
     }
 
     let stackJoined = false;
 
-    if (exifInfo !== false) {
+    if (options.exifInfo !== false) {
       stackJoined = true;
       builder = builder
         .leftJoinAndSelect('asset.exifInfo', 'exifInfo')
@@ -655,34 +673,38 @@ export class AssetRepository implements IAssetRepository {
         .leftJoinAndSelect('stack.assets', 'stackedAssets');
     }
 
-    if (albumId) {
+    if (options.albumId) {
-      builder = builder.leftJoin('asset.albums', 'album').andWhere('album.id = :albumId', { albumId });
+      builder = builder.leftJoin('asset.albums', 'album').andWhere('album.id = :albumId', { albumId: options.albumId });
     }
 
-    if (userIds) {
+    if (options.userIds) {
-      builder = builder.andWhere('asset.ownerId IN (:...userIds )', { userIds });
+      builder = builder.andWhere('asset.ownerId IN (:...userIds )', { userIds: options.userIds });
     }
 
-    if (isArchived !== undefined) {
+    if (options.isArchived !== undefined) {
-      builder = builder.andWhere('asset.isArchived = :isArchived', { isArchived });
+      builder = builder.andWhere('asset.isArchived = :isArchived', { isArchived: options.isArchived });
     }
 
-    if (isFavorite !== undefined) {
+    if (options.isFavorite !== undefined) {
-      builder = builder.andWhere('asset.isFavorite = :isFavorite', { isFavorite });
+      builder = builder.andWhere('asset.isFavorite = :isFavorite', { isFavorite: options.isFavorite });
     }
 
-    if (isTrashed !== undefined) {
+    if (options.isTrashed !== undefined) {
-      builder = builder.andWhere(`asset.deletedAt ${isTrashed ? 'IS NOT NULL' : 'IS NULL'}`).withDeleted();
+      builder = builder.andWhere(`asset.deletedAt ${options.isTrashed ? 'IS NOT NULL' : 'IS NULL'}`).withDeleted();
     }
 
-    if (personId !== undefined) {
+    if (options.isDuplicate !== undefined) {
+      builder = builder.andWhere(`asset.duplicateId ${options.isDuplicate ? 'IS NOT NULL' : 'IS NULL'}`);
+    }
+
+    if (options.personId !== undefined) {
       builder = builder
         .innerJoin('asset.faces', 'faces')
         .innerJoin('faces.person', 'person')
-        .andWhere('person.id = :personId', { personId });
+        .andWhere('person.id = :personId', { personId: options.personId });
     }
 
-    if (withStacked) {
+    if (options.withStacked) {
       if (!stackJoined) {
         builder = builder.leftJoinAndSelect('asset.stack', 'stack').leftJoinAndSelect('stack.assets', 'stackedAssets');
       }

server/src/repositories/index.ts

@@ -2,6 +2,7 @@ import { IAccessRepository } from 'src/interfaces/access.interface';
 import { IActivityRepository } from 'src/interfaces/activity.interface';
 import { IAlbumRepository } from 'src/interfaces/album.interface';
 import { IKeyRepository } from 'src/interfaces/api-key.interface';
+import { IAssetDuplicateRepository } from 'src/interfaces/asset-duplicate.interface';
 import { IAssetStackRepository } from 'src/interfaces/asset-stack.interface';
 import { IAssetRepositoryV1 } from 'src/interfaces/asset-v1.interface';
 import { IAssetRepository } from 'src/interfaces/asset.interface';
@@ -56,11 +57,13 @@ import { SystemMetadataRepository } from 'src/repositories/system-metadata.repos
 import { TagRepository } from 'src/repositories/tag.repository';
 import { UserTokenRepository } from 'src/repositories/user-token.repository';
 import { UserRepository } from 'src/repositories/user.repository';
+import { AssetDuplicateRepository } from 'src/repositories/asset-duplicate.repository';
 
 export const repositories = [
   { provide: IActivityRepository, useClass: ActivityRepository },
   { provide: IAccessRepository, useClass: AccessRepository },
   { provide: IAlbumRepository, useClass: AlbumRepository },
+  { provide: IAssetDuplicateRepository, useClass: AssetDuplicateRepository },
   { provide: IAssetRepository, useClass: AssetRepository },
   { provide: IAssetRepositoryV1, useClass: AssetRepositoryV1 },
   { provide: IAssetStackRepository, useClass: AssetStackRepository },

server/src/repositories/job.repository.ts

@@ -64,6 +64,10 @@ export const JOBS_TO_QUEUE: Record<JobName, QueueName> = {
   [JobName.QUEUE_SMART_SEARCH]: QueueName.SMART_SEARCH,
   [JobName.SMART_SEARCH]: QueueName.SMART_SEARCH,
 
+  // duplicate detection
+  [JobName.QUEUE_DUPLICATE_DETECTION]: QueueName.DUPLICATE_DETECTION,
+  [JobName.DUPLICATE_DETECTION]: QueueName.DUPLICATE_DETECTION,
+
   // XMP sidecars
   [JobName.QUEUE_SIDECAR]: QueueName.SIDECAR,
   [JobName.SIDECAR_DISCOVERY]: QueueName.SIDECAR,

server/src/repositories/search.repository.ts

@@ -9,6 +9,8 @@ import { SmartInfoEntity } from 'src/entities/smart-info.entity';
 import { SmartSearchEntity } from 'src/entities/smart-search.entity';
 import { DatabaseExtension } from 'src/interfaces/database.interface';
 import {
+  AssetDuplicateResult,
+  AssetDuplicateSearch,
   AssetSearchOptions,
   FaceEmbeddingSearch,
   FaceSearchResult,
@@ -144,6 +146,39 @@ export class SearchRepository implements ISearchRepository {
     return results;
   }
 
+  @GenerateSql({
+    params: [
+      {
+        embedding: Array.from({ length: 512 }, Math.random),
+        maxDistance: 0.6,
+        userIds: [DummyValue.UUID],
+      },
+    ],
+  })
+  searchDuplicates({ assetId, maxDistance, userIds }: AssetDuplicateSearch): Promise<AssetDuplicateResult[]> {
+    const cte = this.assetRepository.createQueryBuilder('asset');
+    cte
+      .select('search.assetId', 'assetId')
+      .addSelect('asset.duplicateId', 'duplicateId')
+      .addSelect(`(SELECT embedding FROM smart_search WHERE "assetId" = :assetId) <=> search.embedding`, 'distance')
+      .innerJoin('asset.smartSearch', 'search')
+      .where('asset.ownerId IN (:...userIds )')
+      .andWhere('asset.id != :assetId')
+      .andWhere('asset.isVisible = :isVisible')
+      .orderBy('search.embedding <=> (SELECT embedding FROM smart_search WHERE "assetId" = :assetId)')
+      .limit(64)
+      .setParameters({ assetId, isVisible: true, userIds });
+
+    const builder = this.assetRepository.manager
+      .createQueryBuilder()
+      .addCommonTableExpression(cte, 'cte')
+      .from('cte', 'res')
+      .select('res.*')
+      .where('res.distance <= :maxDistance', { maxDistance });
+
+    return builder.getRawMany() as any as Promise<AssetDuplicateResult[]>;
+  }
+
   @GenerateSql({
     params: [
       {

server/src/services/asset.service.ts

@@ -248,6 +248,11 @@ export class AssetService {
     return data;
   }
 
+  async getDuplicates(auth: AuthDto): Promise<AssetResponseDto[]> {
+    const res = await this.assetRepository.getDuplicates({ userIds: [auth.user.id] });
+    return res.map((a) => mapAsset(a, { auth }));
+  }
+
   async update(auth: AuthDto, id: string, dto: UpdateAssetDto): Promise<AssetResponseDto> {
     await this.access.requirePermission(auth, Permission.ASSET_UPDATE, id);
 

server/src/services/job.service.ts

@@ -115,6 +115,11 @@ export class JobService {
         return this.jobRepository.queue({ name: JobName.QUEUE_SMART_SEARCH, data: { force } });
       }
 
+      case QueueName.DUPLICATE_DETECTION: {
+        await this.configCore.requireFeature(FeatureFlag.SMART_SEARCH);
+        return this.jobRepository.queue({ name: JobName.QUEUE_DUPLICATE_DETECTION, data: { force } });
+      }
+
       case QueueName.METADATA_EXTRACTION: {
         return this.jobRepository.queue({ name: JobName.QUEUE_METADATA_EXTRACTION, data: { force } });
       }
@@ -194,7 +199,11 @@ export class JobService {
   }
 
   private isConcurrentQueue(name: QueueName): name is ConcurrentQueueName {
-    return ![QueueName.FACIAL_RECOGNITION, QueueName.STORAGE_TEMPLATE_MIGRATION].includes(name);
+    return ![
+      QueueName.FACIAL_RECOGNITION,
+      QueueName.STORAGE_TEMPLATE_MIGRATION,
+      QueueName.DUPLICATE_DETECTION,
+    ].includes(name);
   }
 
   async handleNightlyJobs() {
@@ -296,6 +305,13 @@ export class JobService {
         break;
       }
 
+      case JobName.SMART_SEARCH: {
+        if (item.data.source === 'upload') {
+          await this.jobRepository.queue({ name: JobName.DUPLICATE_DETECTION, data: item.data });
+        }
+        break;
+      }
+
       case JobName.USER_DELETION: {
         this.eventRepository.clientBroadcast(ClientEvent.USER_DELETE, item.data.id);
         break;

server/src/services/microservices.service.ts

@@ -14,6 +14,7 @@ import { StorageService } from 'src/services/storage.service';
 import { SystemConfigService } from 'src/services/system-config.service';
 import { UserService } from 'src/services/user.service';
 import { otelSDK } from 'src/utils/instrumentation';
+import { SearchService } from './search.service';
 
 @Injectable()
 export class MicroservicesService {
@@ -31,6 +32,7 @@ export class MicroservicesService {
     private storageService: StorageService,
     private userService: UserService,
     private databaseService: DatabaseService,
+    private searchService: SearchService,
   ) {}
 
   async init() {
@@ -47,6 +49,8 @@ export class MicroservicesService {
       [JobName.USER_SYNC_USAGE]: () => this.userService.handleUserSyncUsage(),
       [JobName.QUEUE_SMART_SEARCH]: (data) => this.smartInfoService.handleQueueEncodeClip(data),
       [JobName.SMART_SEARCH]: (data) => this.smartInfoService.handleEncodeClip(data),
+      [JobName.QUEUE_DUPLICATE_DETECTION]: (data) => this.searchService.handleQueueSearchDuplicates(data),
+      [JobName.DUPLICATE_DETECTION]: (data) => this.searchService.handleSearchDuplicates(data),
       [JobName.STORAGE_TEMPLATE_MIGRATION]: () => this.storageTemplateService.handleMigration(),
       [JobName.STORAGE_TEMPLATE_MIGRATION_SINGLE]: (data) => this.storageTemplateService.handleMigrationSingle(data),
       [JobName.QUEUE_MIGRATION]: () => this.mediaService.handleQueueMigration(),

server/src/services/search.service.ts

@@ -17,16 +17,29 @@ import {
 } from 'src/dtos/search.dto';
 import { AssetOrder } from 'src/entities/album.entity';
 import { AssetEntity } from 'src/entities/asset.entity';
-import { IAssetRepository } from 'src/interfaces/asset.interface';
+import { IAssetDuplicateRepository } from 'src/interfaces/asset-duplicate.interface';
+import { IAssetRepository, WithoutProperty } from 'src/interfaces/asset.interface';
+import { ICryptoRepository } from 'src/interfaces/crypto.interface';
+import {
+  IBaseJob,
+  IEntityJob,
+  IJobRepository,
+  JOBS_ASSET_PAGINATION_SIZE,
+  JobName,
+  JobStatus,
+} from 'src/interfaces/job.interface';
 import { IMachineLearningRepository } from 'src/interfaces/machine-learning.interface';
 import { IMetadataRepository } from 'src/interfaces/metadata.interface';
 import { IPartnerRepository } from 'src/interfaces/partner.interface';
 import { IPersonRepository } from 'src/interfaces/person.interface';
 import { ISearchRepository, SearchExploreItem, SearchStrategy } from 'src/interfaces/search.interface';
 import { ISystemConfigRepository } from 'src/interfaces/system-config.interface';
+import { ImmichLogger } from 'src/utils/logger';
+import { usePagination } from 'src/utils/pagination';
 
 @Injectable()
 export class SearchService {
+  private logger = new ImmichLogger(SearchService.name);
   private configCore: SystemConfigCore;
 
   constructor(
@@ -37,6 +50,9 @@ export class SearchService {
     @Inject(IAssetRepository) private assetRepository: IAssetRepository,
     @Inject(IPartnerRepository) private partnerRepository: IPartnerRepository,
     @Inject(IMetadataRepository) private metadataRepository: IMetadataRepository,
+    @Inject(ICryptoRepository) private cryptoRepository: ICryptoRepository,
+    @Inject(IAssetDuplicateRepository) private assetDuplicateRepository: IAssetDuplicateRepository,
+    @Inject(IJobRepository) private jobRepository: IJobRepository,
   ) {
     this.configCore = SystemConfigCore.create(configRepository);
   }
@@ -139,6 +155,84 @@ export class SearchService {
     }
   }
 
+  async handleQueueSearchDuplicates({ force }: IBaseJob): Promise<JobStatus> {
+    const { machineLearning } = await this.configCore.getConfig();
+    if (!machineLearning.enabled || !machineLearning.clip.enabled) {
+      return JobStatus.SKIPPED;
+    }
+
+    const assetPagination = usePagination(JOBS_ASSET_PAGINATION_SIZE, (pagination) => {
+      return force
+        ? this.assetRepository.getAll(pagination, { isVisible: true })
+        : this.assetRepository.getWithout(pagination, WithoutProperty.DUPLICATE);
+    });
+
+    for await (const assets of assetPagination) {
+      await this.jobRepository.queueAll(
+        assets.map((asset) => ({ name: JobName.DUPLICATE_DETECTION, data: { id: asset.id } })),
+      );
+    }
+
+    return JobStatus.SUCCESS;
+  }
+
+  async handleSearchDuplicates({ id }: IEntityJob): Promise<JobStatus> {
+    const { machineLearning } = await this.configCore.getConfig();
+
+    const asset = await this.assetRepository.getById(id, { smartSearch: true });
+    if (!asset) {
+      this.logger.error(`Asset ${id} not found`);
+      return JobStatus.FAILED;
+    }
+
+    if (!asset.isVisible) {
+      this.logger.debug(`Asset ${id} is not visible, skipping`);
+      return JobStatus.SKIPPED;
+    }
+
+    if (asset.duplicateId) {
+      this.logger.debug(`Asset ${id} already has a duplicateId, skipping`);
+      return JobStatus.SKIPPED;
+    }
+
+    if (!asset.resizePath) {
+      this.logger.debug(`Asset ${id} is missing preview image`);
+      return JobStatus.FAILED;
+    }
+
+    if (!asset.smartSearch?.embedding) {
+      this.logger.debug(`Asset ${id} is missing embedding`);
+      return JobStatus.FAILED;
+    }
+
+    const duplicateAssets = await this.searchRepository.searchDuplicates({
+      assetId: asset.id,
+      embedding: asset.smartSearch.embedding,
+      maxDistance: machineLearning.clip.duplicateThreshold,
+      userIds: [asset.ownerId],
+    });
+
+    const assetIds = [asset.id];
+
+    if (duplicateAssets.length > 0) {
+      this.logger.debug(
+        `Found ${duplicateAssets.length} duplicate${duplicateAssets.length === 1 ? '' : 's'} for asset ${asset.id}`,
+      );
+
+      const duplicateIds = duplicateAssets.map((duplicate) => duplicate.duplicateId).filter(Boolean);
+      const duplicateId = duplicateIds[0] || this.cryptoRepository.randomUUID();
+
+      assetIds.push(...duplicateAssets.map((duplicate) => duplicate.assetId));
+
+      await this.assetDuplicateRepository.upsert(duplicateId, assetIds, duplicateIds);
+    }
+
+    const duplicatesDetectedAt = new Date();
+    await this.assetRepository.upsertJobStatus(assetIds.map((assetId) => ({ assetId, duplicatesDetectedAt })));
+
+    return JobStatus.SUCCESS;
+  }
+
   // TODO: remove after implementing new search filters
   /** @deprecated */
   async search(auth: AuthDto, dto: SearchDto): Promise<SearchResponseDto> {

server/src/services/system-config.service.spec.ts

@@ -74,6 +74,7 @@ const updatedConfig = Object.freeze<SystemConfig>({
     clip: {
       enabled: true,
       modelName: 'ViT-B-32__openai',
+      duplicateThreshold: 0.03,
     },
     facialRecognition: {
       enabled: true,

web/src/lib/components/admin-page/jobs/jobs-panel.svelte

@@ -88,6 +88,12 @@
       subtitle: 'Run machine learning on assets to support smart search',
       disabled: !$featureFlags.smartSearch,
     },
+    [JobName.DuplicateDetection]: {
+      icon: mdiImageSearch,
+      title: getJobName(JobName.DuplicateDetection),
+      subtitle: 'Run machine learning on assets to detect near-duplicate images',
+      disabled: !$featureFlags.smartSearch,
+    },
     [JobName.FaceDetection]: {
       icon: mdiFaceRecognition,
       title: getJobName(JobName.FaceDetection),

web/src/lib/components/shared-components/side-bar/side-bar.svelte

@@ -131,6 +131,19 @@
       </svelte:fragment>
     </SideBarLink>
 
+    <SideBarLink title="Duplicates" routeId="/(user)/duplicates" icon={mdiArchiveArrowDownOutline}>
+      <svelte:fragment slot="moreInformation">
+        {#await getStats({ isArchived: true })}
+          <LoadingSpinner />
+        {:then data}
+          <div>
+            <p>{data.videos.toLocaleString($locale)} Videos</p>
+            <p>{data.images.toLocaleString($locale)} Photos</p>
+          </div>
+        {/await}
+      </svelte:fragment>
+    </SideBarLink>
+
     {#if $featureFlags.trash}
       <SideBarLink title="Trash" routeId="/(user)/trash" icon={mdiTrashCanOutline}>
         <svelte:fragment slot="moreInformation">

web/src/lib/constants.ts

@@ -20,6 +20,7 @@ export enum AppRoute {
   ALBUMS = '/albums',
   LIBRARIES = '/libraries',
   ARCHIVE = '/archive',
+  DUPLICATES = '/duplicates',
   FAVORITES = '/favorites',
   PEOPLE = '/people',
   PLACES = '/places',

web/src/lib/utils.ts

@@ -115,6 +115,7 @@ export const getJobName = (jobName: JobName) => {
     [JobName.MetadataExtraction]: 'Extract Metadata',
     [JobName.Sidecar]: 'Sidecar Metadata',
     [JobName.SmartSearch]: 'Smart Search',
+    [JobName.DuplicateDetection]: 'Duplicate Detection',
     [JobName.FaceDetection]: 'Face Detection',
     [JobName.FacialRecognition]: 'Facial Recognition',
     [JobName.VideoConversion]: 'Transcode Videos',

web/src/routes/(user)/duplicates/+page.svelte

@@ -0,0 +1,21 @@
+<script lang="ts">
+  import GalleryViewer from '$lib/components/shared-components/gallery-viewer/gallery-viewer.svelte';
+  import { getAssetDuplicates, type AssetResponseDto } from '@immich/sdk';
+  import type { Viewport } from '$lib/stores/assets.store';
+  import { onMount } from 'svelte';
+
+  let assets: AssetResponseDto[] = [];
+  const viewport: Viewport = { width: 0, height: 0 };
+
+  onMount(async () => {
+    assets = await getAssetDuplicates();
+  });
+</script>
+
+<section
+  class="relative mb-12 bg-immich-bg dark:bg-immich-dark-bg m-4"
+  bind:clientHeight={viewport.height}
+  bind:clientWidth={viewport.width}
+>
+  <GalleryViewer {assets} {viewport}></GalleryViewer>
+</section>

web/src/routes/(user)/duplicates/+page.ts

@@ -0,0 +1,12 @@
+import { authenticate } from '$lib/utils/auth';
+import type { PageLoad } from './$types';
+
+export const load = (async () => {
+  await authenticate();
+
+  return {
+    meta: {
+      title: 'Duplicates',
+    },
+  };
+}) satisfies PageLoad;

web/src/routes/(user)/duplicates/photos/[assetId]/+page.ts

@@ -0,0 +1,7 @@
+import { AppRoute } from '$lib/constants';
+import { redirect } from '@sveltejs/kit';
+import type { PageLoad } from './$types';
+
+export const load: PageLoad = () => {
+  redirect(302, AppRoute.DUPLICATES);
+};