extractor :ocr score below 80% reject doc

src/main/kotlin/com/android/trisolarisserver/component/DocumentExtractionService.kt

@@ -22,6 +22,13 @@ class DocumentExtractionService(
 
     fun extractAndApply(localImageUrl: String, publicImageUrl: String, document: GuestDocument, propertyId: UUID): ExtractionResult {
         val results = linkedMapOf<String, String>()
+        val ocrResult = paddleOcrClient.extract(document.storagePath)
+        if (ocrResult?.rejected == true) {
+            results["docType"] = "REJECTED"
+            results["rejectReason"] = "LOW_OCR_SCORE"
+            results["ocrAverage"] = ocrResult.averageScore?.toString() ?: "UNKNOWN"
+            return ExtractionResult(results, false)
+        }
         val detections = listOf(
             Detection(
                 detect = {
@@ -89,7 +96,7 @@ class DocumentExtractionService(
                         for ((key, question) in aadharFrontQuestions) {
                             results[key] = llamaClient.ask(localImageUrl, question)
                         }
-                        ensureAadhaarId(localImageUrl, publicImageUrl, document, results)
+                        ensureAadhaarId(localImageUrl, publicImageUrl, document, results, ocrResult)
                     }
                 }
             ),
@@ -278,7 +285,8 @@ class DocumentExtractionService(
         localImageUrl: String,
         publicImageUrl: String,
         document: GuestDocument,
-        results: MutableMap<String, String>
+        results: MutableMap<String, String>,
+        ocrResult: PaddleOcrResult?
     ) {
         val key = DocumentPrompts.ID_NUMBER.first
         val current = cleanedValue(results[key])
@@ -296,7 +304,6 @@ class DocumentExtractionService(
             results[key] = formatAadhaar(retryNormalized)
             return
         }
-        val ocrResult = paddleOcrClient.extract(document.storagePath)
         if (ocrResult != null) {
             val ocrCandidate = ocrResult.aadhaar
             if (ocrCandidate != null && isValidAadhaar(ocrCandidate.replace(" ", ""))) {

src/main/kotlin/com/android/trisolarisserver/component/PaddleOcrClient.kt

@@ -22,7 +22,9 @@ class PaddleOcrClient(
     @Value("\${ocr.paddle.baseUrl:https://ocr.hoteltrisolaris.in}")
     private val baseUrl: String,
     @Value("\${ocr.paddle.minScore:0.9}")
-    private val minScore: Double
+    private val minScore: Double,
+    @Value("\${ocr.paddle.minAverageScore:0.8}")
+    private val minAverageScore: Double
 ) {
     private val logger = LoggerFactory.getLogger(PaddleOcrClient::class.java)
 
@@ -33,9 +35,11 @@ class PaddleOcrClient(
 
         return try {
             val output = callOcr(path)
+            val average = averageScore(output.scores)
             val filtered = filterByScore(output.texts, output.scores, minScore)
             val aadhaar = extractAadhaar(filtered)
-            PaddleOcrResult(filtered, aadhaar)
+            val rejected = average != null && average < minAverageScore
+            PaddleOcrResult(filtered, aadhaar, average, rejected)
         } catch (ex: Exception) {
             logger.warn("PaddleOCR failed: {}", ex.message)
             null
@@ -71,6 +75,11 @@ class PaddleOcrClient(
         }
     }
 
+    private fun averageScore(scores: List<Double>): Double? {
+        if (scores.isEmpty()) return null
+        return scores.sum() / scores.size
+    }
+
     private fun extractAadhaar(texts: List<String>): String? {
         val joined = texts.joinToString(" ")
         val candidates = mutableListOf<String>()
@@ -88,7 +97,9 @@ class PaddleOcrClient(
 
 data class PaddleOcrResult(
     val texts: List<String>,
-    val aadhaar: String?
+    val aadhaar: String?,
+    val averageScore: Double?,
+    val rejected: Boolean
 )
 
 private data class OcrPayload(

src/main/resources/application.properties

@@ -29,4 +29,6 @@ ai.llama.minP=0.2
 ai.llama.repeatPenalty=1.0
 ai.llama.topK=40
 ocr.paddle.enabled=true
+ocr.paddle.baseUrl=https://ocr.hoteltrisolaris.in/
 ocr.paddle.minScore=0.9
+ocr.paddle.minAverageScore=0.8