49 docs tagged with "NeoMind"

Get NeoMind running in 5 minutes: install → configure LLM → connect first device → see data on dashboard → ask AI Chat. Every step has a checkpoint, tips, and troubleshooting.

NeoMind AI Agent guide: autonomous agent concepts, execution modes (Focused/Free), scheduling (cron/event/interval), resource binding, memory system, execution history, and status management.

A complete solution for person and object detection with notification push using the NeoMind platform, offering two implementation paths — YOLO Inference extension (edge inference) and AI Agent (LLM-powered analysis) — supporting NE101/NE301 smart cameras.

NeoMind AI Chat guide: query and control devices in natural language, create dashboards and rules, upload images for visual analysis, tool call mechanism, Chat vs Agent, and multi-session management.

Build NeoMind extensions, dashboard components, and contribute to the main project efficiently using Claude Code and AI-assisted development: project context setup, extension workflow, component workflow, prompt patterns, and real-world examples.

ne101_camera component architecture deep-dive: 5-layer decomposition of the 1972-line IIFE (helper / template / sub-component / main / export), 3 exported components (NE101CameraPanel / ConfigPanel / AdvancedPanel), the WebSocket-priority + REST-fallback data flow, the ROI overlay pipeline, and an architectural comparison with metric_card

NeoMind automation rules guide: rule structure, conditions (comparison/range/logical), actions (notify/execute/trigger_agent), triggers (data_change/schedule/manual), with UI walkthrough, CLI, API examples, and import/export.

CamThink NE101 sensing camera product capabilities, why a dedicated component is needed, its device-bound positioning in the NeoMind ecosystem, and the progression from metric_card

Engineering case studies — 7 selected real-world cases from 17 extensions + 6 components, with full engineering analysis

ne101_camera component build: IIFE injection pattern (window.React + jsxRuntime), named export object, five-layer module structure, three React-in-IIFE pitfalls (#310 hook order, frozen input, conditional useState), ConfigPanel/AdvancedPanel/ExtDropdown sub-components, shadcn CSS class replica

Configure LLM backends in NeoMind: local Ollama (recommended qwen3.5:4b) and cloud models (OpenAI/Anthropic/Qwen/DeepSeek/GLM etc.) — setup steps, CLI commands, model selection, and multimodal capability.

Contributing to the NeoMind main project: development setup, code standards (Rust clippy / frontend ESLint), commit conventions, CI pipeline, PR workflow, and testing requirements.

NeoMind system architecture overview: process model, data lifecycle, data transforms, extension mechanism, agent model, and notification system. Based on actual codebase, user-facing.

NeoMind dashboard component development guide: ZIP package structure, complete manifest.json reference, bundle.js IIFE format, component Props API, CSS variable theming, data source binding, complete Temperature Gauge example, installation and debugging.

ne101_camera data contract: MQTT device telemetry, extension response normalization (boxes_x1y1x2y2 / objects_bbox / detections_bbox / ocr_text_blocks), JSON string detections parsing, ROI Sutherland-Hodgman clipping algorithm, virtual metrics output prefix mapping

NeoMind Data Push guide: push device telemetry to external Webhook or MQTT Broker in real time or on a schedule, with target configuration, data filtering, retry strategy, batch delivery, delivery logs and stats.

NeoMind data transforms guide: use JavaScript code to transform device telemetry in real time, generate derived metrics, invoke extension commands, with transform builder UI, scopes, testing, CLI/API management, and import/export.

ne101_camera deep dive: 133-commit version evolution, the rise and fall of Transform lifecycle debug traces (4 debug commits finally cleaned up by 00a59cc), the Boa engine console.log crash incident, the _configHash performance optimization, source hygiene recap (3 files, zero backups)

A multi-stream RTSP video analytics solution on NG4500 (NVIDIA Jetson Orin) via the deepstream extension — using NVIDIA DeepStream 7.1 to run hardware-accelerated inference and tracking on 2–32 RTSP cameras on a single NG4500, emitting line-crossing, ROI-intrusion, and counting business events and re-publishing annotated video streams to NeoMind. NeoMind runs on a Mac/PC; the DeepStream sidecar runs on the NG4500 and communicates over a remote bridge.

NeoMind developer guide overview: start from one of four repo dimensions (device types / extensions / dashboard components / main project), with tech stack, crate layout, and entry points to deeper docs.

NeoMind device type development guide: device data model, DeviceTypeTemplate, ConnectionConfig, MQTT topics and webhook data formats, auto-discovery flow, CLI device management, ESP32/Python hands-on examples.

End-to-end guide for building a NeoMind extension from scratch with neomind-extension-sdk: scaffolding, Cargo.toml, Extension trait implementation, neomind_export!, cross-platform compile, .nep packaging, install & debug.

NeoMind extension management guide: install/uninstall extensions (.nep), official marketplace, extension details (overview/configuration/commands/metrics/logs), extension capabilities (metrics/commands/components), process isolation and crash protection, CLI and REST API.

neomind-extension-sdk reference: Extension trait, ExtensionMetadata, MetricDescriptor, neomind_export! FFI macro, capability declaration, ML model lifecycle (lazy-load + keep-loaded), cross-platform packaging (cdylib + panic=unwind).

ne101_camera extension-side contract: the processingExtensionId generic AI processing paradigm, the AI_EXT_IDS whitelist, the EXT_MODES mode catalog (imageArg/responseType/command triple), the __imageData injection mechanism, the locate-anything-v2 NMS threshold special-case, and the extension degradation fallback

A face recognition solution based on the NeoMind platform, using the Face Recognition extension to detect faces and identify individuals, with real-time dashboard display, history review, and AI Chat natural language queries for NE101/NE301 smart cameras.

ne101_camera frontend consumption: detections fetching, JSON string parsing, per-class coloring (golden-angle HSV), SVG overlay rendering (polygon + rect fallback), object-cover coordinate transform, ResizeObserver callback-ref pattern, Transform tiered lifecycle

NeoMind core glossary: definitions, relationships, and examples for Device, Device Type, Extension, Capability, Metric, DataSourceId, Agent, AI Chat, Memory, Transform, Rule, Cooldown, Dashboard, Widget, LLM Backend, MQTT Broker, Telemetry, SSE, and more.

Complete install flow for NeoMind on desktop (macOS/Windows/Linux) or server, covering one-line script, Docker, manual install, nginx reverse proxy, and development setup.

ne101_camera integration tests: end-to-end test matrix (test_bundle.js 35KB), ROI overlay verification (Sutherland-Hodgman clipping + object-cover mapping), multi-extension switching tests (locate-anything-v2 / image-analyzer-v2 / yolo-device-inference / ocr-device-inference), source_ts alignment verification, WS+REST dual-channel tests

An open-vocabulary visual grounding solution via the LocateAnything extension — based on the LocateAnything-3B vision-language model, it finds / counts arbitrary objects in a frame by natural language, with phrase grounding, text localization, GUI grounding, and pointing. Supports zero-shot detection and counting for classes a fixed detector doesn't know or ad-hoc needs.

Write your first NeoMind dashboard component — a complete engineering walkthrough of metric_card (IIFE injection + OKLCH glass + multi-format normalization)

Demonstrates the AI processing pipeline (processingExtensionId) of the NeoMind ne101_camera smart-camera component, with a full paddle-ocr-v6 OCR walkthrough — automatically feeding NE101/NE301 captured frames into a local OCR engine, with ROI overlay, virtual-metric write-back, history, and AI Chat queries. The same pipeline also supports object detection, open-vocabulary grounding, and other extensions. Uses water-meter reading and product-label recognition as worked scenarios.

NeoMind Dashboard's most complex device-bound component, the ne101_camera flagship case: 8 subpages covering everything from NE101 device capabilities to the processingExtensionId generic AI processing contract, ROI overlay rendering, and the React-in-IIFE engineering pattern

NeoMind notifications and messages complete guide: configure 9 message channels (Webhook, Email, Telegram, WeCom, DingTalk, Slack, Feishu), channel filters, message lifecycle, CLI and REST API.

A NeoMind-based OCR solution for general text recognition, using the OCR extension to extract text from images with dashboard display, history viewing, and AI Chat natural language queries, supporting NE101/NE301 smart cameras.

Complete guide to onboarding devices into NeoMind via MQTT (embedded broker, auto-discovery), HTTP webhook, manual registration, or an external broker. Includes ESP32/Python examples and the draft approval flow.

NeoMind standard protocol bridge case study: hand-written WS-Discovery multicast discovery, SOAP/WS-Security PasswordDigest client, PTZ control — no onvif-rs dependency, ~2700 lines of pure Rust covering ONVIF Profile S core capabilities

NeoMind technical architecture deep dive: crate layout and dependencies, main process + extension process isolation, event bus, extension FFI ABI, redb storage layer, Tokio concurrency and semaphores.

NeoMind REST API reference: base URL, auth (JWT + API Key), unified response format, main endpoint groups (devices / dashboards / rules / agents / messages / extensions / data-push / LLM backends), Swagger entry, error format.

Shared engineering standards — centralized reference for metadata/capability/version/build/test/release/security, referenced by all case studies

Hardware, operating system, network port, and runtime requirements for NeoMind desktop and server deployment, including recommended local LLM (Ollama) configurations.

NeoMind troubleshooting: service startup failures, port conflicts, Ollama connection, MQTT issues, multimodal 400 errors, extension crashes, data directory permissions, and log locations.

NeoMind production-verified vendor bridge case study: Uink-RMS e-paper cloud platform bridging, JWT auth chain, Markdown→Image rendering (pulldown-cmark + ab_glyph + imageproc), regional endpoint routing, DisplayEditorCard frontend integration — contrast with 4 onvif-bridge as proprietary vs standard

NeoMind dashboard guide: create and edit dashboards, component library (value card / charts / toggle / image / video / map), real-time data, data source binding, public sharing links, and mobile responsiveness.

Build your first NeoMind data-type extension — full engineering walkthrough of weather-forecast-v2 (HTTP polling + periodic metrics + React frontend)

NeoMind is an edge-deployed AI platform for IoT that runs LLM-powered agents directly on your hardware, connects to devices via MQTT/BLE/Webhook, automates responses through a rule engine, and visualizes everything on real-time dashboards.

Deploy YOLOv8 object detection to edge devices with NeoMind's first AI inference extension — full engineering walkthrough of lazy model loading, ONNX Runtime dylib governance, and capability-based device frame acquisition

NeoMind's most complex streaming extension: Push-mode real-time video processing, YOLOv11 detection, ROI/line-crossing/smart-capture, ffmpeg-next + nokhwa dual backends, cross-platform ONNX Runtime dylib governance, and frontend MJPEG integration — a complete engineering dissection