feat: integrate retail ai suite of loss prevention within edge ai demo studio (#1044) (#601)

Author: gooishin

samples/ai/edge-ai-demo-studio/README.md

@@ -13,18 +13,35 @@ Edge AI Demo Studio is a modern toolkit for deploying, managing, and serving AI
 - **Web-Based Management:** Manage devices, workloads, and users through a modern web interface.
 - **Edge Optimized:** Built for Intel hardware and edge environments.
 - **AI Services:** AI Services that users can use for their applications
-  - Text Generation (LLM)
-  - Text to Speech (TTS)
-  - Speech to Text (STT)
-  - Embedding 
-  - Lipsync
-  - Image Generation
-  - MCP Manager
-  - Wake Word Detection
-- **Samples:** Samples use cases that implements the ai services
-  - Digital Avatar
-  - RAG Chat
-  - [AI Exam Marking](./frontend/src/samples/ai-exam-marking/README.md)
+  - **Text Generation** — Generate coherent text using large language models optimized for Intel hardware.
+  - **Text to Speech** — Natural-sounding speech synthesis with multiple voice options powered by Kokoro TTS.
+  - **Speech to Text** — Real-time speech recognition with low-latency transcription using Whisper models optimized for Intel hardware.
+  - **Speaker Diarization** — Identify and label speakers in audio recordings using pyannote.audio speaker diarization models.
+  - **Text Embedding** — Generate dense vector embeddings for semantic search and RAG pipelines.
+  - **Reranker** — Rescore and rerank documents by relevance for improved search and RAG pipelines.
+  - **Vector Database** — FAISS-based vector storage for knowledge base management, semantic search, and RAG pipelines.
+  - **Lipsync** — Real-time avatar lip-syncing with Wav2Lip, streamed over WebRTC.
+  - **Image Generation** — Generate images from text prompts using diffusion models accelerated with OpenVINO.
+  - **MCP Manager** — Manage Model Context Protocol servers and their tool integrations.
+  - **Wake Word Detection** — Detect custom wake words from microphone input and send webhook notifications on detection events.
+- **Samples:** Sample use cases that implement the AI services
+  - **Digital Avatar** — Interact with an AI-powered avatar that combines real-time video with intelligent conversation.
+  - **Digital Avatar Lite** — A lightweight animated robot avatar that brings conversations to life with responsive movements and expressions.
+  - **RAG Chatbot** — Upload documents and chat with an AI that retrieves relevant context to answer your questions.
+  - **Medical Scribe** — Automatically transcribe and diarize doctor-patient conversations, then generate structured SOAP notes.
+  - **Webcam Capture with VLM** — Demonstrate the integration of webcam capture and Visual Language Model (VLM) for enhanced interaction.
+  - **[AI Exam Marking](./frontend/src/samples/ai-exam-marking/README.md)** — AI-powered exam marking using OCR and LLM to automatically grade test papers from images.
+  - **PowerPoint Translator** — Translate PowerPoint presentations while preserving formatting using AI.
+  - **[Geti Image Classification](./frontend/src/samples/geti-classifier/README.md)** — Classify images using a local Intel Geti deployment and send feedback for continuous model improvement.
+  - **Synthetic Image Generation** — Generate and edit synthetic images from base images in real-time for dataset augmentation.
+  - **Robotics AI** — A demo showcasing the capabilities of Robotics AI, including real-time object detection and manipulation.
+- **Suites:** Curated industry-specific AI solution packages built on Intel Edge AI Suites
+  - [*Manufacturing AI Suite*](https://github.com/open-edge-platform/edge-ai-suites/tree/main/manufacturing-ai-suite) — A comprehensive toolkit for building, deploying, and scaling AI applications in industrial environments. Enables real-time integration with optimized hardware for production workflow automation, workplace safety, defect detection, and asset tracking.
+    - [**Pallet Defect Detection**](https://github.com/open-edge-platform/edge-ai-suites/tree/main/manufacturing-ai-suite/industrial-edge-insights-vision) — Real-time pallet condition monitoring on warehouse video streams using DL Streamer Pipeline Server, OpenVINO inference, and WebRTC streaming.
+  - [*Metro AI Suite*](https://github.com/open-edge-platform/edge-ai-suites/tree/main/metro-ai-suite) — Accelerates application development for edge AI video safety, security, and smart city use cases. Includes OpenVINO™ toolkit, Deep Learning Streamer, and Intel® oneAPI Toolkit for media analytics and AI performance optimization.
+    - [**Image-Based Video Search**](https://github.com/open-edge-platform/edge-ai-suites/tree/main/metro-ai-suite/image-based-video-search) — Near real-time image-based similarity search over live video streams using YOLOv11 object detection, ResNet-50 feature extraction via DL Streamer, and Milvus vector indexing.
+  - [*Retail AI Suite*](https://github.com/open-edge-platform/edge-ai-suites/tree/main/retail-ai-suite) — Accelerates development of edge AI applications for retail environments, enabling intelligent automation for use cases such as self-checkout, loss prevention, and store analytics with optimized Intel hardware and the OpenVINO™ toolkit.
+    - [**Loss Prevention**](https://github.com/intel-retail/loss-prevention/tree/main) — Real-time self-checkout loss prevention using object detection and analytics to identify mis-scans and suspicious activity at the point of sale.
 
 ## Architecture Diagram
 ![Archictecture Diagram](./docs/Architecture.png)

samples/ai/edge-ai-demo-studio/docs/Architecture.png

@@ -1,6 +1,8 @@
 // Copyright (C) 2026 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 
+import { readFileSync } from "node:fs";
+import { join } from "node:path";
 import { FlatCompat } from "@eslint/eslintrc";
 import { defineConfig, globalIgnores } from "eslint/config";
 import nextVitals from "eslint-config-next/core-web-vitals";
@@ -12,6 +14,98 @@ const compat = new FlatCompat({
   baseDirectory: import.meta.dirname,
 });
 
+// ─── Sample modularity: forbid static imports of optional services ──────────
+//
+// Optional service integrations are wired through the generated feature-provider
+// registry (`@/services/_generated/feature-providers`) so the export tool
+// (scripts/export-samples.mjs) can prune their folders with `--no-optional`.
+// The export walks the *static* import graph and force-includes any service
+// folder it reaches — so a sample that statically imports a service it declares
+// `optional` drags that folder back in and silently defeats `--no-optional`.
+// This rule reads each sample's data.ts and flags those imports.
+// See docs/OPTIONAL-SERVICES.md.
+
+const SAMPLES_DIR = join(import.meta.dirname, "src", "samples");
+// Mirrors the dependency parser in scripts/export-samples.mjs (serviceId before
+// role, no nested braces — matches the shape of ServiceDependency literals).
+const DEP_RE =
+  /\{\s*serviceId\s*:\s*['"]([\w-]+)['"][^{}]*?role\s*:\s*['"](required|optional)['"][^{}]*?\}/gs;
+const optionalDepsCache = new Map();
+
+function optionalServiceIds(sampleId) {
+  const cached = optionalDepsCache.get(sampleId);
+  if (cached) return cached;
+  const ids = new Set();
+  try {
+    const src = readFileSync(join(SAMPLES_DIR, sampleId, "data.ts"), "utf8");
+    DEP_RE.lastIndex = 0;
+    let m;
+    while ((m = DEP_RE.exec(src)) !== null) {
+      if (m[2] === "optional") ids.add(m[1]);
+    }
+  } catch {
+    // Folder has no data.ts (not a sample) — nothing to enforce.
+  }
+  optionalDepsCache.set(sampleId, ids);
+  return ids;
+}
+
+const sampleModularityPlugin = {
+  rules: {
+    "no-optional-service-imports": {
+      meta: {
+        type: "problem",
+        docs: {
+          description:
+            "Disallow a sample from statically importing a service it declares as an optional dependency; optional services must be reached only through the generated feature-provider registry so the export tool can prune them with --no-optional.",
+        },
+        schema: [],
+        messages: {
+          optionalImport:
+            "Sample '{{sample}}' statically imports optional service '{{service}}' ('{{source}}'). A static import drags the service folder back into the export graph, so `--no-optional` can no longer prune it. Wire optional services through @/services/_generated/feature-providers (useFeatureProviders) or a shared @/context bridge instead. See docs/OPTIONAL-SERVICES.md.",
+        },
+      },
+      create(context) {
+        const filename = (context.filename ?? context.getFilename()).replace(
+          /\\/g,
+          "/",
+        );
+        const m = filename.match(/\/src\/samples\/([^/]+)\//);
+        if (!m) return {};
+        const sampleId = m[1];
+        // `common` (shared barrels) and `_generated` are not samples and have no
+        // per-sample role context to enforce.
+        if (sampleId === "common" || sampleId === "_generated") return {};
+        const optional = optionalServiceIds(sampleId);
+        if (optional.size === 0) return {};
+
+        const check = (node) => {
+          const value = node?.value;
+          if (typeof value !== "string") return;
+          // Only `@/services/<id>/...` specifiers carry a service folder; shared
+          // entrypoints (_generated, common, types, registry) never match an
+          // optional service id and are therefore allowed.
+          const svc = value.match(/^@\/services\/([^/]+)(?:\/|$)/);
+          if (svc && optional.has(svc[1])) {
+            context.report({
+              node,
+              messageId: "optionalImport",
+              data: { sample: sampleId, service: svc[1], source: value },
+            });
+          }
+        };
+
+        return {
+          ImportDeclaration: (node) => check(node.source),
+          ExportNamedDeclaration: (node) => check(node.source),
+          ExportAllDeclaration: (node) => check(node.source),
+          ImportExpression: (node) => check(node.source),
+        };
+      },
+    },
+  },
+};
+
 const eslintConfig = defineConfig([
   ...nextVitals,
   ...nextTs,
@@ -75,6 +169,54 @@ const eslintConfig = defineConfig([
       "no-console": "off",
     },
   },
+  {
+    files: ["src/services/**/*.{ts,tsx}"],
+    rules: {
+      "no-restricted-imports": [
+        "error",
+        {
+          patterns: [
+            {
+              group: ["@/samples", "@/samples/**", "**/samples/**"],
+              message:
+                "Layering violation: services must not import from the samples tree (samples sit above services). This keeps services independently prunable and avoids a samples↔services cycle. Put shared contracts in @/context, @/hooks, or @/types. See docs/ARCHITECTURE-LAYERS.md and docs/OPTIONAL-SERVICES.md.",
+            },
+          ],
+        },
+      ],
+    },
+  },
+  {
+    files: ["src/engines/**/*.{ts,tsx}"],
+    rules: {
+      "no-restricted-imports": [
+        "error",
+        {
+          patterns: [
+            {
+              group: [
+                "@/samples",
+                "@/samples/**",
+                "**/samples/**",
+                "@/services",
+                "@/services/**",
+                "**/services/**",
+              ],
+              message:
+                "Layering violation: engines are the bottom layer and must not import from the services or samples trees. Keep engines self-contained; put any shared contract in @/components, @/context, @/hooks, @/lib, or @/types instead. See docs/ARCHITECTURE-LAYERS.md.",
+            },
+          ],
+        },
+      ],
+    },
+  },
+  {
+    files: ["src/samples/**/*.{ts,tsx}"],
+    plugins: { "sample-modularity": sampleModularityPlugin },
+    rules: {
+      "sample-modularity/no-optional-service-imports": "error",
+    },
+  },
   {
     settings: {
       react: { version: "19" },

samples/ai/edge-ai-demo-studio/frontend/eslint.config.mjs

@@ -11,5 +11,13 @@
     "src/components/ui/**",
     "src/components/ai-elements/**",
     "src/payload/migrations/**"
+  ],
+  "ignoreDependencies": [
+    "@streamdown/.+",
+    "^prettier$",
+    "^prettier-plugin-tailwindcss$",
+    "^eslint-plugin-prettier$",
+    "^libsql$",
+    "^hast-util-to-jsx-runtime$"
   ]
 }

samples/ai/edge-ai-demo-studio/frontend/knip.json

@@ -67,6 +67,7 @@ const nextConfig: NextConfig = {
   },
   experimental: {
     proxyTimeout: 1000 * proxyTimeoutSetting(),
+    proxyClientMaxBodySize: 500 * 1024 * 1024,
   },
 }