Developer Documentation

Learn how to integrate the complex RF Drone Classification engine into your Software Defined Radio (SDR) scripts and edge nodes.

How Classification Works

Simply upload a raw RF capture file from any Software Defined Radio (SDR) device. Our AI pipeline handles all the heavy lifting — signal processing, feature extraction, and neural inference — and returns a clean JSON result in milliseconds.

1. Upload Signal
Send your raw SDR binary file via the API
2. AI Analysis
Deep learning model analyses the RF spectrum pattern
3. Get Results
Receive drone class, confidence score, and spectrogram
 The API accepts raw binary files containing interleaved float32 I/Q samples — the standard output format of SDR tools like GNU Radio, SDR#, and RTL-SDR.

Credit Balance Operations

REST API Endpoint

POST /api/v1/predict

Request Headers

Header Name Value Type Description
X-API-Key String Your secret API key (e.g. dr_sk_live_xxxx...). Required for authorization.

Request Body (multipart/form-data)

Param Name Type Requirement Description
file Binary File Required Raw SDR file containing complex interleaving float32 I/Q baseband samples: [I0, Q0, I1, Q1, ...].

Response Schema (application/json)

{
  "success": true,
  "predicted_class": "DJI Mavic Pro",
  "confidence": 0.9458,
  "probabilities": {
    "DJI Mavic Pro": 0.9458,
    "DJI Phantom 4": 0.0215,
    "Parrot Bebop 2": 0.0084,
    "Yuneec Typhoon H": 0.0102,
    "DJI Inspire 2": 0.0051,
    "DIY Custom Drone (F450)": 0.0062,
    "Background Noise (No Drone)": 0.0028
  },
  "latency_ms": 7.42,
  "spectrogram_image": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAO...",
  "sample_count": 20480
}