RTSPanda v0.1.1

RTSPanda v0.1.1 Release Notes

Headline

RTSPanda v0.1.1 adds Android no-Docker startup support and introduces an arm64 Pi-mode thermal monitor with banded alerts and API visibility.

Highlights

• Added scripts/android-up.sh for Termux-native startup (no Docker, no root).
• Added thermal monitor package with four thermal bands and hysteresis handling.
• Added thermal_band to /api/v1/system/stats.
• Added Discord system alert dispatch on Hot thermal band entry.
• Updated README and quickstart with Android and release-aligned setup guidance.

What Changed

Android startup path

• New script: scripts/android-up.sh
  • sets RTSPANDA_MODE=pi
  • sets/creates DATA_DIR
  • runs termux-wake-lock when available
  • launches ./rtspanda

Thermal monitoring

• New package: backend/internal/thermal/monitor.go
• Thermal bands:
  • Normal: <45C
  • Warm: 45-54C
  • Hot: 55-64C
  • Critical: >=65C
• Sensor source priority:
  1. /sys/class/thermal/thermal_zone*/temp
  2. /proc/loadavg proxy fallback
  3. disabled mode with warning
• Hysteresis:
  • Critical -> Hot: 5m
  • Hot -> Warm: 5m
  • Warm -> Normal: 3m

Runtime integration

• cmd/rtspanda/main.go now starts thermal monitor when:
  • GOARCH=arm64 and mode is pi, or
  • THERMAL_MONITOR_ENABLED=true
• THERMAL_AUTO_RESUME defaults to false.
• Thermal transition events are logged with severity:
  • Warm: WARN
  • Hot: ERROR
  • Critical: CRITICAL
• On first Hot-band entry, Discord system alerts are sent to cameras with configured webhook URLs.

API change

• GET /api/v1/system/stats now includes:
  • thermal_band (normal, warm, hot, critical)

Validation

• cd backend && go test ./...

Upgrade Notes

• Existing deployments can update in place.
• Android/Termux users can use scripts/android-up.sh for consistent startup.

RTSPanda v0.1.0

RTSPanda v0.1.0 Release Notes

Headline

RTSPanda v0.1.0 hardens stream orchestration and readiness behavior for production camera fleets, and ships a full operator walkthrough that documents every supported setup method.

Highlights

• mediamtx orchestration now defaults to proactive stream startup (sourceOnDemand=false).
• Stream API now returns initializing until HLS is truly reachable.
• Keepalive logic now uses grace and backoff windows instead of tight path repair loops.
• README is now a production-grade setup and operations guide covering all deployment methods.

What Changed

Streaming Pipeline Stability

• Enforced proactive path startup and retained TCP RTSP transport defaults.
• Added idempotent path sync:
  • no path recreation if current source already matches
  • update only when path is missing or source changed
• Reduced reload churn:
  • keepalive repair waits for unhealthy grace period
  • repeat repairs respect backoff window
  • full reload remains reserved for repeated mediamtx API failures

Stream Readiness Gate

• /api/v1/cameras/:id/stream now withholds hls_url until stream status is online.
• New status value initializing indicates stream path exists but playlist is not yet playable.
• Dashboard and camera views updated to handle initializing cleanly without premature playback attempts.

Documentation

• Rewrote README into a full production walkthrough:
  • Standard full stack
  • Pi mode
  • Pi + remote worker
  • Standalone AI worker host
  • Viewer mode (binary and Docker)
  • Source development setup
  • hardening, validation, troubleshooting, and upgrade guidance

Validation

• cd backend && go test ./...
• cd frontend && npm run build

Upgrade Notes

• Existing installs can pull and redeploy normally.
• No breaking route changes; stream endpoint now intentionally returns empty hls_url until playable.

RTSPanda v0.0.9

RTSPanda v0.0.9 Release Notes

Headline

RTSPanda v0.0.9 formalizes deployment modes as a runtime contract and adds a dedicated Snapshot Intelligence Engine for Raspberry Pi deployments.

Highlights

• Added first-class runtime modes: pi, standard, and viewer.
• Added Pi Snapshot AI pipeline with Claude/OpenAI providers and structured detection event output.
• Added strict Pi deployment guardrails: local YOLO AI-worker paths are explicitly blocked on ARM.
• Rewrote architecture and README docs around mode capabilities and constraints.

What Changed

Runtime and Backend

• Added internal/mode for mode detection, capability gating, and startup banners.
• Updated startup flow in backend/cmd/rtspanda/main.go:
  • YOLO detection manager starts only in mode-capable environments.
  • Pi and Viewer still expose detection endpoints through degraded manager handles.
  • Snapshot AI manager starts in Pi-capable mode when enabled.
• Added CaptureFrameToPath export in internal/detections/capture.go for snapshot pipeline reuse.

Pi AI and Deployment

• Added internal/snapshotai/manager.go and internal/snapshotai/providers.go for interval frame analysis through external vision APIs.
• Updated docker-compose.yml:
  • Explicit RTSPANDA_MODE defaults.
  • Pi profile Snapshot AI environment variables.
• Updated scripts/pi-up.sh:
  • Removed unsupported full Pi path.
  • Added ARM checks to block standalone AI-worker mode on Pi.
  • Improved user-facing deployment and AI option guidance.

Documentation

• Reworked README.md to explain three deployment modes and expected behavior.
• Expanded AI planning docs (AI/ARCHITECTURE.md, AI/DECISIONS.md, AI/PROJECT_CONTEXT.md) to align implementation with mode decisions.

Validation

• cd backend && go test ./...

Upgrade Notes

• Existing docker compose up --build -d server installs remain supported and map to Standard mode.
• Pi installs should use PI_DEPLOYMENT_MODE=pi with optional Snapshot AI env vars.
• Remote AI-worker deployments remain supported via separate server-class hosts.

RTSPanda v0.0.8

RTSPanda v0.0.8 Release Notes

Headline

RTSPanda v0.0.8 turns the project into a more serious edge-video platform: the default single-machine workflow stays intact, while Raspberry Pi deployments gain a cleaner first-run path, deterministic ONNX-only AI builds, and an easy upgrade path to distributed inference.

Highlights

• Added a Pi-first lightweight deployment mode for rtspanda without local AI-worker overhead.
• Added cluster mode so a Pi can ingest streams locally and send frames to a remote AI worker on a second machine.
• Removed AI-worker export fallback from Docker builds. The worker now uses prebuilt ONNX models only.
• Added additive Compose profiles for full, pi, and ai-worker without breaking the existing docker compose up --build -d flow.
• Rewrote setup and deployment docs to clearly support Standard, Pi Standalone, and Pi + AI topologies.
• Fixed Pi standalone startup so it no longer traverses the local AI-worker build path.
• Repaired the default remote ONNX pin to a live Ultralytics asset and improved failure messages for invalid model sources.

What Changed

Platform and Deployment

• docker-compose.yml now supports:
  • standard full-stack deployment
  • lightweight Pi deployment
  • standalone remote AI-worker deployment
• New docker-compose.standalone.yml overlay isolates Pi-only and AI-worker-only deployments from the unprofiled full stack.
• scripts/pi-up.sh now supports:
  • PI_DEPLOYMENT_MODE=pi
  • PI_DEPLOYMENT_MODE=full
  • PI_DEPLOYMENT_MODE=ai-worker
• Standalone launch paths now use build followed by up --no-build so targeted deployments only build the intended service.
• scripts/pi-preflight.sh now checks deployment mode and model-source expectations more accurately for Docker-first Pi workflows.

AI Runtime

• ai_worker/Dockerfile now resolves models deterministically:
  • local prebuilt ONNX file first
  • explicit YOLO_MODEL_URL second
  • named Ultralytics ONNX asset fallback last
• Default remote model pin now targets yolo11n on Ultralytics v8.3.0, which is published and buildable today.
• No PyTorch install path
• No YOLO(...).export(...)
• No runtime model conversion on ARM

Backend Detection Routing

• Added additive AI-target resolution using:
  • AI_MODE=local|remote
  • AI_WORKER_URL=http://<host>:8090
  • DETECTOR_URL as the highest-precedence override
• Detection health now reports AI mode and remote worker context.

Release Quality Improvements

• Refreshed the frontend lockfile so current tests/tooling install cleanly.
• Fixed clean-checkout frontend embed compilation behavior in the backend.

Setup Paths

Standard

docker compose up --build -d

Pi Standalone

./scripts/pi-up.sh

Pi + AI

AI host:

docker compose -f docker-compose.yml -f docker-compose.standalone.yml --profile ai-worker build ai-worker-standalone
docker compose -f docker-compose.yml -f docker-compose.standalone.yml --profile ai-worker up -d --no-build ai-worker-standalone

Pi host:

export AI_WORKER_URL="http://192.168.1.50:8090"
./scripts/pi-up.sh

Validation Checklist

• docker compose config -q
• cd backend && go test ./internal/...
• cd frontend && npm run test -- --config vitest.config.ts
• cd ai_worker && python -m pytest -q

Upgrade Notes

• Existing standard users can continue using docker compose up --build -d.
• Pi users who want the old all-in-one behavior can use PI_DEPLOYMENT_MODE=full ./scripts/pi-up.sh.
• For custom ONNX assets, place model.onnx at repo root or ai_worker/model/model.onnx, then set MODEL_SOURCE=local.

v0.0.7

RTSPanda v0.0.7 Release Notes

Highlights

• Added token-based authentication across backend and frontend, including login/session handling and protected API routes.
• Added initial automated test foundation for backend, frontend, and AI worker.
• Improved Raspberry Pi deployment with architecture-aware container builds and helper scripts.
• Resolved streaming configuration drift and documented latency/stability tuning controls.
• Expanded AI worker runtime safeguards for Pi CPU-only environments.

Included Areas

• Security/auth: backend/internal/auth/, backend/internal/api/router.go, frontend/src/auth/
• Streaming/runtime: backend/internal/streams/mediamtx.go, mediamtx/mediamtx.yml.tmpl, docker-compose.yml
• Deployment: Dockerfile, ai_worker/Dockerfile, scripts/pi-*.sh
• AI runtime: ai_worker/app/main.py, ai_worker/requirements.txt
• Testing/docs: docs/testing-strategy.md, frontend/vitest.config.ts, ai_worker/tests/*

Validation Checklist

• docker compose config -q
• cd backend && go test ./internal/...
• cd frontend && npm run test -- --config vitest.config.ts
• cd ai_worker && python -m pytest -q

Upgrade Notes

• If authentication is enabled, define AUTH_TOKEN before startup.
• For Raspberry Pi, prefer 64-bit OS and use ./scripts/pi-up.sh for first deploy.

v0.0.6

RTSPanda v0.0.6 - Performance, Observability, System Monitoring, Multi-View UX

Tag: v0.0.6
Diff: v0.0.5...v0.0.6

---

Highlights

• Dashboard stream status now uses a shared mediamtx path cache and batch status endpoint.
• API responses under /api/ now support gzip compression.
• New observability surface: request logging, Prometheus-compatible /metrics, and system stats API.
• New readiness endpoint for deeper health checks: GET /api/v1/health/ready.
• Multi-view UX improvements: inline add-camera card + per-panel remove button.
• Settings now includes a live System tab for uptime, memory, goroutines, request count, and network counters.

---

Backend and API Changes

• Added request body limit (256 KB) for camera create/update payloads:
  • backend/internal/api/cameras.go
• Router and middleware updates:
  • backend/internal/api/router.go
  • Added API gzip wrapping, request logging wrapper, /metrics, and system routes.
• New API endpoints:
  • GET /api/v1/cameras/stream-status
  • GET /api/v1/system/stats
  • GET /api/v1/health/ready
  • GET /metrics
• New observability implementation files:
  • backend/internal/api/metrics.go
  • backend/internal/api/middleware.go
  • backend/internal/api/sysinfo.go

---

Streams and Performance

• Added stream path-list cache with 3-second TTL (shared across callers):
  • backend/internal/streams/cache.go
• Stream manager now exposes:
  • StreamStatusMap(cameraIDs []string)
  • IsReady() bool
• Stream keepalive now uses cached path list, reducing mediamtx API pressure.
• mediamtx config now enables native metrics:
  • metrics: yes
  • metricsAddress: 127.0.0.1:9998

---

Database Changes

• Added migration:
  • backend/internal/db/migrations/008_cameras_index.sql
• Index added:
  • idx_cameras_order on cameras(position, created_at)

---

Frontend Changes

• Dashboard now batch-loads camera list + stream status map in parallel:
  • frontend/src/pages/Dashboard.tsx
  • frontend/src/api/cameras.ts
  • frontend/src/components/CameraGrid.tsx
  • frontend/src/components/CameraCard.tsx
• App now uses route-level lazy loading and suspense fallback:
  • frontend/src/App.tsx
• Multi-view enhancements:
  • Inline add-camera card picker
  • Per-panel quick remove (✕)
  • frontend/src/pages/MultiCameraView.tsx
• New System tab + API integration:
  • frontend/src/pages/Settings.tsx
  • frontend/src/api/sysinfo.ts

---

AI Docs and Filesystem Review

Reviewed and updated AI planning/handoff/docs to match this release:

• Updated:
  • AI/CURRENT_FOCUS.md
  • AI/DECISIONS.md
  • AI/HANDOFF.md
  • AI/TODO.md
• Added:
  • AI/AGENTIC-PLATFORM-EXPANSION-GUIDE.md

These docs now reflect:

• v0.0.6 shipped scope
• deferred platform expansion roadmap
• explicit risks and next-priority items

---

User Documentation

• Updated README.md to v0.0.6 summary and compare link.
• Updated human/USER_GUIDE.md with system monitoring and metrics coverage.

---

Validation

• Backend:
  • cd backend && go build ./... (pass)
• Frontend:
  • cd frontend && npm run build (pass)

RTSPanda v0.0.5

RTSPanda v0.0.5 — Frigate Alert Provider + In-App Guides

Tag: v0.0.5
Diff: v0.0.4...v0.0.5

---

Highlights

• Added provider-based Discord detection alerts per camera:
  • yolo (RTSPanda built-in YOLOv8)
  • frigate (external Frigate event ingestion)
• Added Frigate webhook endpoint:
  • POST /api/v1/frigate/events
• Added new in-app Guides page with practical setup walkthroughs:
  • Lorex NVR port-forwarding
  • Tailscale setup
  • Lorex RTSP URL retrieval from login/web UI flow
• Added Support the Developer links to https://248tech.com/donate.

---

Backend / API Updates

• Added Frigate ingestion handler:
  • backend/internal/api/frigate.go
  • Parses Frigate webhook payloads and forwards detection alerts to matching cameras.
  • Matches cameras by provider frigate and frigate_camera_name (or camera name fallback).
  • Ignores non-new Frigate event types.
• Registered new route:
  • POST /api/v1/frigate/events
• Extended notifier API contract to support external detection event dispatch.
• Updated Discord notification flow:
  • Added source-aware detection notifications (YOLOv8 vs Frigate labels in payloads).
  • Ensured YOLO-origin notifications are skipped for cameras configured with provider frigate.
• Added optional Frigate snapshot fetch support via env:
  • FRIGATE_BASE_URL
  • If configured, event snapshots are fetched and attached to Discord alerts when available.

---

Database / Camera Model Updates

• Added migration:
  • backend/internal/db/migrations/007_discord_detection_provider.sql
• New camera fields:
  • discord_detection_provider (default yolo)
  • frigate_camera_name (default empty)
• Updated camera model, service validation, and repository persistence/scan logic:
  • backend/internal/cameras/model.go
  • backend/internal/cameras/service.go
  • backend/internal/cameras/repository.go

---

Frontend Updates

• Added new page:
  • frontend/src/pages/Guides.tsx
• Added sidebar navigation route for Guides:
  • frontend/src/App.tsx
• Added donate/support links in UI:
  • Sidebar support icon
  • Guides page support button
• Extended camera API types for provider fields:
  • frontend/src/api/cameras.ts
• Updated camera setup form:
  • Provider selection (YOLOv8 or Frigate)
  • Optional Frigate camera name
  • Source-aware trigger wording
  • frontend/src/components/CameraForm.tsx
• Updated settings and camera badges to reflect provider-aware alerting:
  • frontend/src/pages/Settings.tsx
  • frontend/src/pages/CameraView.tsx

---

Documentation Updates

• Updated README release section and feature/config/API notes:
  • README.md
• Updated user guide for provider-based alerts and Frigate endpoint/env:
  • human/USER_GUIDE.md

---

Validation

• Backend checks:
  • go test ./... (pass)
• Frontend checks:
  • npm run build (pass)

v0.0.4

RTSPanda v0.0.4 - YOLO UI Polishing, Security, UI, Reliability, Connectivity

Tag: v0.0.4
Diff: v0.0.3...v0.0.4

YOLO UI Polishing

• Added per-camera ignore-zone editing using polygon drawing from live camera frames.
• Added ignore-zone rendering in camera view video overlays.
• Improved YOLO camera configuration UX and error guidance in forms.
• Improved detection feedback in camera view actions and status messages.

Security Updates

• Added strict backend validation for ignore-zone polygon structure and coordinates.
• Added validation and normalization for integrations settings input.
• Added centralized OpenAI settings service with key-set/clear handling.
• Added provider-specific validation rules for external storage configuration.

UI Updates

• Added multi-camera view (up to 4 cameras) with batch screenshot and Discord actions.
• Added Settings -> Integrations UI for OpenAI captions and external storage sync.
• Updated app shell navigation to include multi-view and extension download access.
• Added dashboard Reset Network control for stream recovery workflows.
• Added bundled Chrome PiP extension package and install instructions.

Reliability Updates

• Migrated AI worker runtime from PyTorch/ultralytics to ONNX Runtime + numpy.
• Added multi-stage AI worker Docker build with ONNX export at build time.
• Added stream keepalive health checks (mediamtx API + HLS reachability).
• Added per-camera stream reset and full-stream reload API endpoints.
• Added async manual Discord recording trigger flow with improved backend logging.
• Added conservative Docker memory/runtime defaults for constrained hosts.

Connectivity

• Added external video storage sync service for:
  • Local Server (NAS/SMB/NFS path)
  • Dropbox (rclone)
  • Google Drive (rclone)
  • OneDrive (rclone)
  • Proton Drive (rclone)
• Added app settings API endpoints:
  • GET /api/v1/settings
  • PUT /api/v1/settings
• Added stream control API endpoints:
  • POST /api/v1/cameras/{id}/stream/reset
  • POST /api/v1/streams/reset

Docs Updated

• README.md (v0.0.4 summary + config updates)
• docs/EXTERNAL_VIDEO_STORAGE.md
• extensions/chrome-rtspanda-pip/README.md
• frontend/public/downloads/rtspanda-chrome-pip-extension-install.md

RTSPanda v0.0.3

RTSPanda v0.0.3 — YOLO Reliability + Discord Media Controls

This release improves detector stability and expands Discord alerting/media workflows.

---

Highlights

• Improved YOLO detector reliability in Docker deployments.
• Added verbose backend + AI worker logging around detection requests/results.
• Added per-camera Discord trigger controls for detection and interval screenshots.
• Added manual camera-view actions:
  • Screenshot to Discord
  • Record to Discord
• Added richer clip/media handling with format fallback:
  • webm, webp, gif
• Updated user/docs wording to YOLO-first alerting, while keeping legacy alert-rule APIs for compatibility.

---

Detection Reliability Fixes

• Added FFmpeg RTSP timeout option fallback for older builds:
  • try -rw_timeout
  • fallback to -timeout
  • fallback to no timeout flag
• Added detector URL fallback behavior in backend detection client.
• Improved detector request failure aggregation/logging across fallback URLs.
• Added runtime dependencies to the AI worker image to prevent startup/runtime crashes in container environments.
• Ensured multipart image uploads include an explicit image content type expected by the detector.

---

New Camera Discord Controls

Per-camera fields added:

• discord_trigger_on_detection
• discord_trigger_on_interval
• discord_screenshot_interval_seconds
• discord_include_motion_clip
• discord_motion_clip_seconds
• discord_record_format
• discord_record_duration_seconds

Database migration:

• backend/internal/db/migrations/005_discord_triggers.sql

---

New API Endpoints

• POST /api/v1/cameras/{id}/discord/screenshot
• POST /api/v1/cameras/{id}/discord/record

---

Notes

• Legacy alert-rule webhook APIs remain available for external workflows.
• Retention cleanup, delivery retry/backoff, and event pagination remain on the post-release backlog.

v0.0.2

Full Changelog: v0.0.1...v0.0.2