feat(ocr-pipeline): add Step 4 row detection UI in admin frontend

Insert rows step between columns and words in the pipeline wizard.
Shows overlay image, row list with type badges, and ground truth controls.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Benjamin Admin
2026-02-28 01:28:05 +01:00
parent 04b83d5f46
commit b58aecd081
3 changed files with 300 additions and 7 deletions

View File

@@ -6,6 +6,7 @@ import { PipelineStepper } from '@/components/ocr-pipeline/PipelineStepper'
import { StepDeskew } from '@/components/ocr-pipeline/StepDeskew'
import { StepDewarp } from '@/components/ocr-pipeline/StepDewarp'
import { StepColumnDetection } from '@/components/ocr-pipeline/StepColumnDetection'
import { StepRowDetection } from '@/components/ocr-pipeline/StepRowDetection'
import { StepWordRecognition } from '@/components/ocr-pipeline/StepWordRecognition'
import { StepCoordinates } from '@/components/ocr-pipeline/StepCoordinates'
import { StepReconstruction } from '@/components/ocr-pipeline/StepReconstruction'
@@ -142,10 +143,11 @@ export default function OcrPipelinePage() {
1: 'Begradigung',
2: 'Entzerrung',
3: 'Spalten',
4: 'Woerter',
5: 'Koordinaten',
6: 'Rekonstruktion',
7: 'Validierung',
4: 'Zeilen',
5: 'Woerter',
6: 'Koordinaten',
7: 'Rekonstruktion',
8: 'Validierung',
}
const renderStep = () => {
@@ -157,12 +159,14 @@ export default function OcrPipelinePage() {
case 2:
return <StepColumnDetection sessionId={sessionId} onNext={handleNext} />
case 3:
return <StepWordRecognition />
return <StepRowDetection sessionId={sessionId} onNext={handleNext} />
case 4:
return <StepCoordinates />
return <StepWordRecognition />
case 5:
return <StepReconstruction />
return <StepCoordinates />
case 6:
return <StepReconstruction />
case 7:
return <StepGroundTruth />
default:
return null

View File

@@ -28,6 +28,7 @@ export interface SessionInfo {
deskew_result?: DeskewResult
dewarp_result?: DewarpResult
column_result?: ColumnResult
row_result?: RowResult
}
export interface DeskewResult {
@@ -91,10 +92,35 @@ export interface ManualColumnDivider {
export type ColumnTypeKey = PageRegion['type']
export interface RowResult {
rows: RowItem[]
summary: Record<string, number>
total_rows: number
duration_seconds: number
}
export interface RowItem {
index: number
x: number
y: number
width: number
height: number
word_count: number
row_type: 'content' | 'header' | 'footer'
gap_before: number
}
export interface RowGroundTruth {
is_correct: boolean
corrected_rows?: RowItem[]
notes?: string
}
export const PIPELINE_STEPS: PipelineStep[] = [
{ id: 'deskew', name: 'Begradigung', icon: '📐', status: 'pending' },
{ id: 'dewarp', name: 'Entzerrung', icon: '🔧', status: 'pending' },
{ id: 'columns', name: 'Spalten', icon: '📊', status: 'pending' },
{ id: 'rows', name: 'Zeilen', icon: '📏', status: 'pending' },
{ id: 'words', name: 'Woerter', icon: '🔤', status: 'pending' },
{ id: 'coordinates', name: 'Koordinaten', icon: '📍', status: 'pending' },
{ id: 'reconstruction', name: 'Rekonstruktion', icon: '🏗️', status: 'pending' },

View File

@@ -0,0 +1,263 @@
'use client'
import { useCallback, useEffect, useState } from 'react'
import type { RowResult, RowGroundTruth } from '@/app/(admin)/ai/ocr-pipeline/types'
const KLAUSUR_API = '/klausur-api'
interface StepRowDetectionProps {
sessionId: string | null
onNext: () => void
}
export function StepRowDetection({ sessionId, onNext }: StepRowDetectionProps) {
const [rowResult, setRowResult] = useState<RowResult | null>(null)
const [detecting, setDetecting] = useState(false)
const [error, setError] = useState<string | null>(null)
const [gtNotes, setGtNotes] = useState('')
const [gtSaved, setGtSaved] = useState(false)
useEffect(() => {
if (!sessionId) return
const fetchSession = async () => {
try {
const res = await fetch(`${KLAUSUR_API}/api/v1/ocr-pipeline/sessions/${sessionId}`)
if (res.ok) {
const info = await res.json()
if (info.row_result) {
setRowResult(info.row_result)
return
}
}
} catch (e) {
console.error('Failed to fetch session info:', e)
}
// No cached result — run auto
runAutoDetection()
}
fetchSession()
// eslint-disable-next-line react-hooks/exhaustive-deps
}, [sessionId])
const runAutoDetection = useCallback(async () => {
if (!sessionId) return
setDetecting(true)
setError(null)
try {
const res = await fetch(`${KLAUSUR_API}/api/v1/ocr-pipeline/sessions/${sessionId}/rows`, {
method: 'POST',
})
if (!res.ok) {
const err = await res.json().catch(() => ({ detail: res.statusText }))
throw new Error(err.detail || 'Zeilenerkennung fehlgeschlagen')
}
const data: RowResult = await res.json()
setRowResult(data)
} catch (e) {
setError(e instanceof Error ? e.message : 'Unbekannter Fehler')
} finally {
setDetecting(false)
}
}, [sessionId])
const handleGroundTruth = useCallback(async (isCorrect: boolean) => {
if (!sessionId) return
const gt: RowGroundTruth = {
is_correct: isCorrect,
notes: gtNotes || undefined,
}
try {
await fetch(`${KLAUSUR_API}/api/v1/ocr-pipeline/sessions/${sessionId}/ground-truth/rows`, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify(gt),
})
setGtSaved(true)
} catch (e) {
console.error('Ground truth save failed:', e)
}
}, [sessionId, gtNotes])
if (!sessionId) {
return (
<div className="flex flex-col items-center justify-center py-16 text-center">
<div className="text-5xl mb-4">📏</div>
<h3 className="text-lg font-medium text-gray-700 dark:text-gray-300 mb-2">
Schritt 4: Zeilenerkennung
</h3>
<p className="text-gray-500 dark:text-gray-400 max-w-md">
Bitte zuerst Schritte 1-3 abschliessen.
</p>
</div>
)
}
const overlayUrl = `${KLAUSUR_API}/api/v1/ocr-pipeline/sessions/${sessionId}/image/rows-overlay`
const dewarpedUrl = `${KLAUSUR_API}/api/v1/ocr-pipeline/sessions/${sessionId}/image/dewarped`
const rowTypeColors: Record<string, string> = {
header: 'bg-gray-200 dark:bg-gray-600 text-gray-700 dark:text-gray-300',
content: 'bg-blue-100 dark:bg-blue-900/30 text-blue-700 dark:text-blue-300',
footer: 'bg-gray-200 dark:bg-gray-600 text-gray-700 dark:text-gray-300',
}
return (
<div className="space-y-4">
{/* Loading */}
{detecting && (
<div className="flex items-center gap-2 text-teal-600 dark:text-teal-400 text-sm">
<div className="animate-spin w-4 h-4 border-2 border-teal-500 border-t-transparent rounded-full" />
Zeilenerkennung laeuft...
</div>
)}
{/* Images: overlay vs clean */}
<div className="grid grid-cols-2 gap-4">
<div>
<div className="text-xs font-medium text-gray-500 dark:text-gray-400 mb-1">
Mit Zeilen-Overlay
</div>
<div className="border rounded-lg overflow-hidden dark:border-gray-700 bg-gray-50 dark:bg-gray-900">
{rowResult ? (
// eslint-disable-next-line @next/next/no-img-element
<img
src={`${overlayUrl}?t=${Date.now()}`}
alt="Zeilen-Overlay"
className="w-full h-auto"
/>
) : (
<div className="aspect-[3/4] flex items-center justify-center text-gray-400 text-sm">
{detecting ? 'Erkenne Zeilen...' : 'Keine Daten'}
</div>
)}
</div>
</div>
<div>
<div className="text-xs font-medium text-gray-500 dark:text-gray-400 mb-1">
Entzerrtes Bild
</div>
<div className="border rounded-lg overflow-hidden dark:border-gray-700 bg-gray-50 dark:bg-gray-900">
{/* eslint-disable-next-line @next/next/no-img-element */}
<img
src={dewarpedUrl}
alt="Entzerrt"
className="w-full h-auto"
/>
</div>
</div>
</div>
{/* Row summary */}
{rowResult && (
<div className="bg-white dark:bg-gray-800 rounded-xl border border-gray-200 dark:border-gray-700 p-4 space-y-3">
<div className="flex items-center justify-between">
<h4 className="text-sm font-medium text-gray-700 dark:text-gray-300">
Ergebnis: {rowResult.total_rows} Zeilen erkannt
</h4>
<span className="text-xs text-gray-400">
{rowResult.duration_seconds}s
</span>
</div>
{/* Type summary badges */}
<div className="flex gap-2">
{Object.entries(rowResult.summary).map(([type, count]) => (
<span
key={type}
className={`px-2 py-0.5 rounded text-xs font-medium ${rowTypeColors[type] || 'bg-gray-100 text-gray-600'}`}
>
{type}: {count}
</span>
))}
</div>
{/* Row list */}
<div className="max-h-64 overflow-y-auto space-y-1">
{rowResult.rows.map((row) => (
<div
key={row.index}
className={`flex items-center gap-3 px-3 py-1.5 rounded text-xs font-mono ${
row.row_type === 'header' || row.row_type === 'footer'
? 'bg-gray-50 dark:bg-gray-700/50 text-gray-500'
: 'text-gray-600 dark:text-gray-400'
}`}
>
<span className="w-8 text-right text-gray-400">R{row.index}</span>
<span className={`px-1.5 py-0.5 rounded text-[10px] uppercase font-semibold ${rowTypeColors[row.row_type] || ''}`}>
{row.row_type}
</span>
<span>y={row.y}</span>
<span>h={row.height}px</span>
<span>{row.word_count} Woerter</span>
{row.gap_before > 0 && (
<span className="text-gray-400">gap={row.gap_before}px</span>
)}
</div>
))}
</div>
</div>
)}
{/* Controls */}
{rowResult && (
<div className="bg-white dark:bg-gray-800 rounded-xl border border-gray-200 dark:border-gray-700 p-4 space-y-3">
<div className="flex items-center gap-3">
<button
onClick={() => runAutoDetection()}
disabled={detecting}
className="px-3 py-1.5 text-xs border rounded-lg hover:bg-gray-50 dark:hover:bg-gray-700 dark:border-gray-600 disabled:opacity-50"
>
Erneut erkennen
</button>
<div className="flex-1" />
{/* Ground truth */}
{!gtSaved ? (
<>
<input
type="text"
placeholder="Notizen (optional)"
value={gtNotes}
onChange={(e) => setGtNotes(e.target.value)}
className="px-2 py-1 text-xs border rounded dark:bg-gray-700 dark:border-gray-600 w-48"
/>
<button
onClick={() => handleGroundTruth(true)}
className="px-3 py-1.5 text-xs bg-green-600 text-white rounded-lg hover:bg-green-700"
>
Korrekt
</button>
<button
onClick={() => handleGroundTruth(false)}
className="px-3 py-1.5 text-xs bg-red-600 text-white rounded-lg hover:bg-red-700"
>
Fehlerhaft
</button>
</>
) : (
<span className="text-xs text-green-600 dark:text-green-400">
Ground Truth gespeichert
</span>
)}
<button
onClick={onNext}
className="px-4 py-1.5 text-xs bg-teal-600 text-white rounded-lg hover:bg-teal-700 font-medium"
>
Weiter
</button>
</div>
</div>
)}
{error && (
<div className="p-3 bg-red-50 dark:bg-red-900/20 text-red-600 dark:text-red-400 rounded-lg text-sm">
{error}
</div>
)}
</div>
)
}