diff --git a/.github/workflows/pr-target-check.yml b/.github/workflows/pr-target-check.yml new file mode 100644 index 00000000..cb7ee50f --- /dev/null +++ b/.github/workflows/pr-target-check.yml @@ -0,0 +1,29 @@ +name: PR Target Check + +on: + pull_request: + types: [opened, reopened, edited, synchronize] + +jobs: + check-target: + runs-on: ubuntu-latest + steps: + - name: Enforce branch strategy + run: | + BASE="${{ github.event.pull_request.base.ref }}" + HEAD="${{ github.event.pull_request.head.ref }}" + + echo "PR: $HEAD → $BASE" + + if [ "$BASE" = "master" ] && [ "$HEAD" != "develop" ]; then + echo "❌ Only the 'develop' branch can open PRs to 'master'." + echo " Please target 'develop' instead." + exit 1 + fi + + if [ "$BASE" = "develop" ] && [ "$HEAD" = "master" ]; then + echo "❌ Do not merge 'master' back into 'develop'." + exit 1 + fi + + echo "✅ PR target is valid: $HEAD → $BASE" diff --git a/README.md b/README.md index 0919465f..5608a232 100644 --- a/README.md +++ b/README.md @@ -27,13 +27,13 @@
-### 🛡️ Introducing [SharpAI Aegis](https://www.sharpai.org) — Desktop App for DeepCamera +### 🚀 Introducing [Aegis AI](https://www.sharpai.org) — The World's First AI Home Security Agent -**Use DeepCamera's AI skills through a desktop app with LLM-powered setup, agent chat, and smart alerts — connected to your mobile via Discord / Telegram / Slack.** +**GPT + Local VLM powered security agent for your Mac and PC. Always connected to your mobile via Discord / Telegram / Slack.** -[SharpAI Aegis](https://www.sharpai.org) is the desktop companion for DeepCamera. It uses LLM to automatically set up your environment, configure camera skills, and manage the full AI pipeline — no manual Docker or CLI required. It also adds an intelligent agent layer: persistent memory, agentic chat with your cameras, AI video generation, voice (TTS), and conversational messaging via Discord / Telegram / Slack. +**Aegis AI** is the next-generation product from SharpAI — a desktop security agent that connects to any camera, understands scenes using Vision Language Models, and talks to your family through the apps they already use. DeepCamera will be integrated into Aegis AI as its detection and model training engine. -[**📦 Download SharpAI Aegis**](https://www.sharpai.org) +[**🌐 Visit www.sharpai.org**](https://www.sharpai.org) · [**📦 Download Aegis AI**](https://github.com/SharpAI/DeepCamera/releases/tag/Aegis)
@@ -41,12 +41,12 @@

Browse & Run VLMs Locally from HuggingFace

-SharpAI Aegis — Browse and download vision language models from HuggingFace +Aegis AI — Browse and download vision language models from HuggingFace

Download and run SmolVLM2, Qwen-VL, LFM2.5, LLaVA, MiniCPM-V directly on your machine. Even a Mac M1 Mini 8GB works.

Chat with your AI Security Agent

-SharpAI Aegis — Ask your agent what happened and get real answers +Aegis AI — Ask your agent what happened and get real answers

Ask "anyone entering the room?" — Aegis searches your footage and gives you a real answer with timestamps and clips.

@@ -54,6 +54,35 @@ --- +### 🏗️ The SharpAI Ecosystem + +| | **DeepCamera** (This Repo) | **Aegis AI** | +|---|---|---| +| **Role** | Open-source AI/ML engine | AI Security Agent | +| **Frontend UI** | No (CLI + Docker) | Full desktop app | +| **What it does** | Object detection, person REID, model training, feature extraction | VLM scene understanding, smart alerts, chat interface, unified timeline | +| **Cameras** | Via Home-Assistant (RTSP, Blink, Nest) | Direct: Ring, Blink, RTSP/ONVIF, webcam, iPhone, Android | +| **AI Models** | YOLOv7, FastReID, custom training | GPT Vision, local VLMs via llama-server, HuggingFace models | +| **LLM / VLM** | No | Yes — local + cloud | +| **Runs on** | Docker on edge devices & desktops | Native desktop app (Mac, Windows, Linux) | +| **Alerts** | No | Slack, Discord, Telegram | +| **Installation** | Docker + CLI setup | One-click download | + +> **DeepCamera** is being integrated into Aegis AI as the detection and model training layer. DeepCamera will remain fully open-source — handling object detection, person re-identification, and custom model training — while Aegis AI provides the agent layer: VLM-powered scene understanding, camera connections, agentic memory, and smart notifications. + +### 🗺️ DeepCamera Roadmap + +DeepCamera is evolving into a full AI skill platform. Planned features: + +- [ ] **Upgrade object detection** to 2026 state-of-the-art YOLO models +- [ ] **VLM analysis backend** — offline scene understanding of recorded clips using vision language models +- [ ] **AI Studios backend** — SAM2 interactive segmentation, DINOv3 grounding, depth estimation, feature extraction +- [ ] **Direct camera provider plugins** — Blink, Ring, Eufy, Reolink, Tapo, RTSP/ONVIF (beyond Home Assistant) +- [ ] **Messaging channel plugins** — Telegram, Discord, Slack, WhatsApp +- [ ] **Automation triggers** — MQTT, webhooks, Home Assistant events +- [ ] **go2rtc streaming** — RTSP to WebRTC live view +- [x] **Skill architecture** — pluggable `SKILL.md` interface for all capabilities + --- ## 🎯 Overview diff --git a/skills.json b/skills.json index 2438d590..3920dc5c 100644 --- a/skills.json +++ b/skills.json @@ -15,6 +15,591 @@ "integrations": "Smart home and IoT platform bridges" }, "skills": [ + { + "id": "yolo-detection-2026", + "name": "YOLO Object Detection (2026)", + "description": "State-of-the-art real-time object detection. Person, vehicle, animal, and 80+ COCO class detection.", + "version": "1.0.0", + "category": "detection", + "path": "skills/detection/yolo-detection-2026", + "tags": [ + "yolo", + "detection", + "person", + "vehicle", + "real-time" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "gpu": "optional", + "ram_gb": 2 + }, + "capabilities": [ + "live_detection" + ], + "ui_unlocks": [ + "detection_overlay", + "alert_config" + ] + }, + { + "id": "dinov3-grounding", + "name": "DINOv3 Visual Grounding", + "description": "Open-vocabulary detection — describe what to find in natural language, DINOv3 locates it.", + "version": "1.0.0", + "category": "detection", + "path": "skills/detection/dinov3-grounding", + "tags": [ + "dinov3", + "grounding", + "open-vocabulary", + "features" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64" + ], + "requirements": { + "python": ">=3.9", + "gpu": "recommended", + "ram_gb": 4 + }, + "capabilities": [ + "live_detection", + "feature_extraction" + ], + "ui_unlocks": [ + "dinov3_studio", + "annotation_grounding" + ] + }, + { + "id": "person-recognition", + "name": "Person Recognition (ReID)", + "description": "Track and re-identify individuals across cameras using appearance embeddings.", + "version": "1.0.0", + "category": "detection", + "path": "skills/detection/person-recognition", + "tags": [ + "reid", + "person", + "tracking", + "identification" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64" + ], + "requirements": { + "python": ">=3.9", + "gpu": "optional", + "ram_gb": 2 + }, + "capabilities": [ + "live_detection" + ], + "ui_unlocks": [ + "identity_gallery", + "reappearance_alerts" + ] + }, + { + "id": "vlm-scene-analysis", + "name": "VLM Scene Analysis", + "description": "Offline scene understanding using vision language models — describe what happened in clips.", + "version": "1.0.0", + "category": "analysis", + "path": "skills/analysis/vlm-scene-analysis", + "tags": [ + "vlm", + "scene", + "description", + "offline", + "search" + ], + "platforms": [ + "linux-x64", + "darwin-arm64", + "darwin-x64" + ], + "requirements": { + "python": ">=3.9", + "gpu": "recommended", + "ram_gb": 4 + }, + "capabilities": [ + "offline_analysis" + ], + "ui_unlocks": [ + "analysis_badges", + "search_by_description" + ] + }, + { + "id": "sam2-segmentation", + "name": "SAM2 Interactive Segmentation", + "description": "Click-to-segment using Meta's Segment Anything 2 — pixel-perfect masks for annotation.", + "version": "1.0.0", + "category": "analysis", + "path": "skills/analysis/sam2-segmentation", + "tags": [ + "sam2", + "segmentation", + "masks", + "annotation" + ], + "platforms": [ + "linux-x64", + "darwin-arm64", + "darwin-x64" + ], + "requirements": { + "python": ">=3.9", + "gpu": "recommended", + "ram_gb": 4 + }, + "capabilities": [ + "live_transform" + ], + "ui_unlocks": [ + "sam2_studio", + "annotation_segmentation" + ] + }, + { + "id": "depth-estimation", + "name": "Depth Estimation", + "description": "Real-time monocular depth maps using Depth Anything v2.", + "version": "1.0.0", + "category": "transformation", + "path": "skills/transformation/depth-estimation", + "tags": [ + "depth", + "3d", + "spatial", + "overlay" + ], + "platforms": [ + "linux-x64", + "darwin-arm64", + "darwin-x64" + ], + "requirements": { + "python": ">=3.9", + "gpu": "recommended", + "ram_gb": 3 + }, + "capabilities": [ + "live_transform" + ], + "ui_unlocks": [ + "depth_studio" + ] + }, + { + "id": "dataset-annotation", + "name": "Dataset Annotation", + "description": "AI-assisted dataset creation — bbox, SAM2, DINOv3 annotation with COCO export.", + "version": "1.0.0", + "category": "annotation", + "path": "skills/annotation/dataset-annotation", + "tags": [ + "annotation", + "coco", + "dataset", + "labeling", + "training" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "gpu": "optional", + "ram_gb": 2 + }, + "capabilities": [ + "annotation" + ], + "ui_unlocks": [ + "annotation_view", + "dataset_explorer", + "training_pipeline" + ] + }, + { + "id": "camera-provider-eufy", + "name": "Eufy Cameras", + "description": "Eufy camera integration — local RTSP streaming and event clips.", + "version": "1.0.0", + "category": "camera-providers", + "path": "skills/camera-providers/eufy", + "tags": [ + "eufy", + "anker", + "camera", + "rtsp", + "local" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "clip_feed", + "discover_cameras", + "live_stream" + ], + "ui_unlocks": [ + "camera_timeline", + "clip_feed", + "live_view" + ] + }, + { + "id": "camera-provider-reolink", + "name": "Reolink Cameras", + "description": "Reolink camera integration — RTSP, HTTP API, ONVIF.", + "version": "1.0.0", + "category": "camera-providers", + "path": "skills/camera-providers/reolink", + "tags": [ + "reolink", + "camera", + "rtsp", + "onvif", + "poe" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "clip_feed", + "discover_cameras", + "live_stream", + "snapshot" + ], + "ui_unlocks": [ + "camera_timeline", + "clip_feed", + "live_view" + ] + }, + { + "id": "camera-provider-tapo", + "name": "Tapo Cameras", + "description": "TP-Link Tapo camera integration — RTSP and ONVIF.", + "version": "1.0.0", + "category": "camera-providers", + "path": "skills/camera-providers/tapo", + "tags": [ + "tapo", + "tplink", + "camera", + "rtsp", + "onvif" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "clip_feed", + "discover_cameras", + "live_stream" + ], + "ui_unlocks": [ + "camera_timeline", + "clip_feed", + "live_view" + ] + }, + { + "id": "go2rtc-cameras", + "name": "go2rtc Multi-Camera Streaming", + "description": "Register RTSP cameras for low-latency WebRTC live view via go2rtc.", + "version": "1.0.0", + "category": "streaming", + "path": "skills/streaming/go2rtc-cameras", + "tags": [ + "go2rtc", + "rtsp", + "webrtc", + "streaming", + "live" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "live_stream" + ], + "ui_unlocks": [ + "live_grid_view", + "webrtc_player" + ] + }, + { + "id": "channel-matrix", + "name": "Matrix Channel", + "description": "Matrix/Element messaging channel for Clawdbot agent.", + "version": "1.0.0", + "category": "channels", + "path": "skills/channels/matrix", + "tags": [ + "matrix", + "element", + "messaging", + "chat" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "messaging_channel" + ], + "ui_unlocks": [ + "channel_config" + ] + }, + { + "id": "channel-line", + "name": "LINE Channel", + "description": "LINE messaging channel for Clawdbot agent.", + "version": "1.0.0", + "category": "channels", + "path": "skills/channels/line", + "tags": [ + "line", + "messaging", + "chat", + "asia" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "messaging_channel" + ], + "ui_unlocks": [ + "channel_config" + ] + }, + { + "id": "channel-signal", + "name": "Signal Channel", + "description": "Signal messaging channel for Clawdbot agent — most private option.", + "version": "1.0.0", + "category": "channels", + "path": "skills/channels/signal", + "tags": [ + "signal", + "messaging", + "chat", + "privacy" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "messaging_channel" + ], + "ui_unlocks": [ + "channel_config" + ] + }, + { + "id": "mqtt-automation", + "name": "MQTT Automation", + "description": "Publish Aegis events to MQTT broker for Home Assistant, Node-RED, etc.", + "version": "1.0.0", + "category": "automation", + "path": "skills/automation/mqtt", + "tags": [ + "mqtt", + "automation", + "homeassistant", + "nodered" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "automation" + ], + "ui_unlocks": [ + "automation_rules" + ] + }, + { + "id": "webhook-trigger", + "name": "Webhook Trigger", + "description": "POST Aegis events to webhook URLs — Zapier, IFTTT, Make, n8n.", + "version": "1.0.0", + "category": "automation", + "path": "skills/automation/webhook", + "tags": [ + "webhook", + "automation", + "zapier", + "ifttt" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "automation" + ], + "ui_unlocks": [ + "automation_rules" + ] + }, + { + "id": "ha-automation-trigger", + "name": "HA Automation Trigger", + "description": "Fire Aegis detection events in Home Assistant for automations.", + "version": "1.0.0", + "category": "automation", + "path": "skills/automation/ha-trigger", + "tags": [ + "homeassistant", + "automation", + "events" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "automation" + ], + "ui_unlocks": [ + "automation_rules" + ] + }, + { + "id": "homeassistant-bridge", + "name": "Home Assistant Bridge", + "description": "Bidirectional HA integration — HA cameras in, AI detection results out.", + "version": "1.0.0", + "category": "integrations", + "path": "skills/integrations/homeassistant-bridge", + "tags": [ + "homeassistant", + "bridge", + "camera", + "detection" + ], + "platforms": [ + "linux-x64", + "linux-arm64", + "darwin-arm64", + "darwin-x64", + "win-x64" + ], + "requirements": { + "python": ">=3.9", + "ram_gb": 1 + }, + "capabilities": [ + "ha_bridge" + ], + "ui_unlocks": [ + "ha_entity_browser", + "ha_status_panel" + ] + }, { "id": "home-security-benchmark", "name": "Home Security AI Benchmark",