StagesDetect
SignatureDetector
YoloOnnxDetector pre-set to StabRise/signature_detection, writing a `signatures` column.
- Import
import { SignatureDetector } from '@stabrise/scaledp/detect'- Group
- Detect
- Reads
- image
- Writes
- boxes
- Needs
onnxruntime-web
import { Pipeline } from '@stabrise/scaledp'
import { PdfToImage } from '@stabrise/scaledp/pdf'
import { SignatureDetector } from '@stabrise/scaledp/detect'
import { ImageDrawBoxes } from '@stabrise/scaledp'
const rows = await new Pipeline([
new PdfToImage({ resolution: 200 }),
new SignatureDetector(),
new ImageDrawBoxes({ inputCols: ['image', 'signatures'], outputCol: 'annotated' }),
]).transform(file)Identical to YoloOnnxDetector with
model: 'StabRise/signature_detection', labels: ['signature'],
outputCol: 'signatures' and outputType: 'signature'. Every parameter is still
overridable.
Validators do not run on this subclass
It spreads its pre-set options straight into super rather than through
resolveParams, so the base class's validators are skipped. Passing
model: '' explicitly is accepted here and fails later at init() with a less
useful message.
Parameters
| Parameter | Type | Default | Meaning |
|---|---|---|---|
model | string | 'StabRise/signature_detection' | Hugging Face repo id, or a URL when self-hosting. No default. |
labels | string[] | ['signature'] | Class index → label. Empty falls back to class_<n>. |
scoreThreshold | number 0–1 | 0.2 | Drop detections below this confidence. |
iouThreshold | number 0–1 | 0.5 | Overlap above which two same-class boxes count as duplicates. |
padding | number 0–1 | 0 | Grow each box by this fraction of its size, to avoid clipping edges. |