Vision Platform
@rosepetal/node-red-contrib-vision-platform provides 8 nodes for ML inference, Firebase/Firestore integration, and dataset management in Node-RED.
Installation
bash
cd ~/.node-red
NPM_TOKEN="$(gcloud auth print-access-token)" npm install @rosepetal/node-red-contrib-vision-platformRequirements
- Node.js >= 16, Node-RED >= 3.0.0
- Docker (for inference nodes)
- Firebase project (for dataset nodes)
- NVIDIA GPU + CUDA (optional, for Triton Inferencer)
Nodes
Configuration
| Node | Purpose |
|---|---|
| firebase-config | Central Firebase authentication and connection management with health checks |
Dataset Management
| Node | Purpose |
|---|---|
| dataset-upload | Batch upload images to Firebase with concurrent processing and PNG compression |
| from-dataset | Stream images from Firebase datasets with tag and set filtering |
| list-dataset | Browse and monitor available datasets with real-time sync |
| list-model | Browse available AI models with task normalization |
Inference
| Node | Purpose |
|---|---|
| inferencer | AI inference for detection, classification, segmentation, and anomaly detection (YOLO, Paddle, RF-DETR, TensorRT) |
| ocr-inferencer | Optical character recognition with PaddleOCR (detection, recognition, document rotation) |
| triton-inferencer | TensorRT-optimized inference via NVIDIA Triton Inference Server |
Supported Models
| Framework | Tasks |
|---|---|
| YOLO | Detection, Segmentation, Classification |
| PaddlePaddle | Detection, OCR, Recognition, Document Rotation |
| RF-DETR | Detection |
| TensorRT | Detection, Segmentation, Classification |
| Anomalib | Anomaly Detection |