From c79c34895c7a5cac3ae134e569d27546d10889ff Mon Sep 17 00:00:00 2001 From: Shuming Liu Date: Fri, 7 Aug 2026 16:06:43 +0800 Subject: [PATCH] fixed a bug --- .../example/studentfaceregistry/ImageUtils.kt | 4 +- .../studentfaceregistry/MainActivity.kt | 79 +++++++++- .../data/StudentRepository.kt | 19 ++- .../studentfaceregistry/face/FaceAligner.kt | 125 +++++++++++++++- .../studentfaceregistry/face/FaceEmbedder.kt | 67 +++++++-- .../studentfaceregistry/face/FaceProcessor.kt | 53 +++++-- .../face/SmartEnrollment.kt | 63 ++++++-- .../upload/UploadServer.kt | 135 ++++++++++++++---- 8 files changed, 470 insertions(+), 75 deletions(-) diff --git a/app/src/main/java/com/example/studentfaceregistry/ImageUtils.kt b/app/src/main/java/com/example/studentfaceregistry/ImageUtils.kt index 0e714e3..6814bba 100644 --- a/app/src/main/java/com/example/studentfaceregistry/ImageUtils.kt +++ b/app/src/main/java/com/example/studentfaceregistry/ImageUtils.kt @@ -8,10 +8,10 @@ import androidx.camera.core.ImageProxy @ExperimentalGetImage object ImageUtils { - fun imageProxyToBitmap(imageProxy: ImageProxy): Bitmap { + fun imageProxyToBitmap(imageProxy: ImageProxy, applyRotation: Boolean = true): Bitmap { val image = imageProxy.image ?: error("ImageProxy does not contain an image.") val bitmap = yuv420ToBitmap(image) - if (imageProxy.imageInfo.rotationDegrees == 0) { + if (!applyRotation || imageProxy.imageInfo.rotationDegrees == 0) { return bitmap } val matrix = Matrix().apply { diff --git a/app/src/main/java/com/example/studentfaceregistry/MainActivity.kt b/app/src/main/java/com/example/studentfaceregistry/MainActivity.kt index 5d39e34..1d5490a 100644 --- a/app/src/main/java/com/example/studentfaceregistry/MainActivity.kt +++ b/app/src/main/java/com/example/studentfaceregistry/MainActivity.kt @@ -42,6 +42,7 @@ import com.example.studentfaceregistry.face.MatchType import com.example.studentfaceregistry.face.RecognitionResult import com.google.mlkit.vision.common.InputImage import com.google.mlkit.vision.face.Face +import com.google.mlkit.vision.face.FaceLandmark import com.example.studentfaceregistry.ui.DetectionUi import com.example.studentfaceregistry.ui.OverlayView import com.example.studentfaceregistry.upload.UploadServer @@ -115,6 +116,20 @@ class MainActivity : AppCompatActivity() { repository.students.collectLatest { list -> students = list countText.text = getString(R.string.registered_student_count, list.size) + if (list.isNotEmpty()) { + val first = list.first() + val earliest = list.minOfOrNull { it.createdAt } ?: 0L + val latest = list.maxOfOrNull { it.createdAt } ?: 0L + Log.i( + TAG, + "已加载学生 ${list.size} 人，" + + "特征维度=${first.embedding.size}，" + + "注册时间范围=${fmtTime(earliest)} ~ ${fmtTime(latest)}，" + + "特征范数=${String.format(java.util.Locale.US, "%.3f", embeddingNorm(first.embedding))}" + ) + } else { + Log.i(TAG, "已加载学生 0 人（请先注册）") + } } } @@ -589,8 +604,18 @@ class MainActivity : AppCompatActivity() { if (facesNeedingRefresh.isNotEmpty()) { val refreshStart = SystemClock.elapsedRealtime() val bitmap = ImageUtils.imageProxyToBitmap(image) + var unrotatedBitmap: Bitmap? = null facesNeedingRefresh.forEach { (index, face) -> - val refreshed = refreshRecognition(faceProcessor, bitmap, face) + var refreshed = refreshRecognition(faceProcessor, bitmap, face) + if (refreshed.matchType == MatchType.NO_MATCH && image.imageInfo.rotationDegrees != 0) { + val fallbackBitmap = unrotatedBitmap ?: ImageUtils + .imageProxyToBitmap(image, applyRotation = false) + .also { unrotatedBitmap = it } + val fallback = refreshRecognition(faceProcessor, fallbackBitmap, face) + if (recognitionScore(fallback) > recognitionScore(refreshed)) { + refreshed = fallback + } + } refreshedResults[index] = refreshed } val refreshMs = SystemClock.elapsedRealtime() - refreshStart @@ -941,6 +966,13 @@ class MainActivity : AppCompatActivity() { bitmap: Bitmap, face: Face ): RecognitionResult { + val landmarkCount = listOf( + face.getLandmark(FaceLandmark.LEFT_EYE), + face.getLandmark(FaceLandmark.RIGHT_EYE), + face.getLandmark(FaceLandmark.NOSE_BASE), + face.getLandmark(FaceLandmark.MOUTH_LEFT), + face.getLandmark(FaceLandmark.MOUTH_RIGHT) + ).count { it != null } val primary = matcher.findNearest(faceProcessor.embedFace(bitmap, face), students) val cached = face.trackingId?.let { recognitionCache[it] } // 未匹配时只做单次推理；仅在疑似匹配或人脸明显移动时才用多裁剪升级， @@ -950,6 +982,7 @@ class MainActivity : AppCompatActivity() { primary.matchType != MatchType.NO_MATCH if (!needsEscalation) { + logRecognitionDebug("仅主候选", landmarkCount, primary) return primary } @@ -960,9 +993,39 @@ class MainActivity : AppCompatActivity() { best = result } } + logRecognitionDebug("含升级候选", landmarkCount, primary, best) return best } + private fun logRecognitionDebug( + stage: String, + landmarkCount: Int, + primary: RecognitionResult, + best: RecognitionResult = primary + ) { + val primaryName = primary.student?.let { "${it.name}(${it.studentNo})" } ?: "无" + val bestName = best.student?.let { "${it.name}(${it.studentNo})" } ?: "无" + Log.i( + TAG, + String.format( + java.util.Locale.US, + "识别 debug: 阶段=%s 关键点=%d/5 主候选=%s(余弦=%.3f 欧氏=%.3f %s) " + + "最优=%s(余弦=%.3f 欧氏=%.3f %s) 学生=%d", + stage, + landmarkCount, + primaryName, + primary.cosineSimilarity, + primary.distance, + primary.matchType, + bestName, + best.cosineSimilarity, + best.distance, + best.matchType, + students.size + ) + ) + } + private fun resolveRecognitionResult( face: Face, refreshedResult: RecognitionResult?, @@ -1020,7 +1083,21 @@ class MainActivity : AppCompatActivity() { } } + private fun embeddingNorm(embedding: FloatArray): Double { + var sum = 0.0 + for (value in embedding) { + sum += value.toDouble() * value.toDouble() + } + return kotlin.math.sqrt(sum) + } + + private fun fmtTime(timestamp: Long): String { + return java.text.SimpleDateFormat("yyyy-MM-dd HH:mm", java.util.Locale.US) + .format(java.util.Date(timestamp)) + } + companion object { + private const val TAG = "MainActivity" private const val MATCHER_PREFS_NAME = "face_matcher_config" private const val KEY_EUCLIDEAN_THRESHOLD = "euclidean_threshold" private const val KEY_COSINE_THRESHOLD = "cosine_threshold" diff --git a/app/src/main/java/com/example/studentfaceregistry/data/StudentRepository.kt b/app/src/main/java/com/example/studentfaceregistry/data/StudentRepository.kt index d265648..4e70a9b 100644 --- a/app/src/main/java/com/example/studentfaceregistry/data/StudentRepository.kt +++ b/app/src/main/java/com/example/studentfaceregistry/data/StudentRepository.kt @@ -13,7 +13,15 @@ class StudentRepository(context: Context) { val students: StateFlow> = _students fun add(student: Student) { - val next = _students.value.toMutableList().apply { add(student) } + val studentToStore = if (student.id == 0L) { + student.copy(id = stableId(student.studentNo)) + } else { + student + } + val next = _students.value + .filterNot { it.studentNo == studentToStore.studentNo } + .toMutableList() + .apply { add(studentToStore) } .sortedByDescending { it.createdAt } _students.value = next save(next) @@ -32,10 +40,11 @@ class StudentRepository(context: Context) { return buildList { for (index in 0 until array.length()) { val obj = array.getJSONObject(index) + val studentNo = obj.getString("studentNo") add( Student( - id = obj.optLong("id", index.toLong()), - studentNo = obj.getString("studentNo"), + id = obj.optLong("id").takeIf { it != 0L } ?: stableId(studentNo), + studentNo = studentNo, name = obj.getString("name"), photoUri = obj.getString("photoUri"), embedding = decodeEmbedding(obj.getJSONArray("embedding")), @@ -80,4 +89,8 @@ class StudentRepository(context: Context) { } } } + + private fun stableId(studentNo: String): Long { + return (studentNo.hashCode().toLong() and 0xffffffffL).coerceAtLeast(1L) + } } diff --git a/app/src/main/java/com/example/studentfaceregistry/face/FaceAligner.kt b/app/src/main/java/com/example/studentfaceregistry/face/FaceAligner.kt index 65d3671..2742981 100644 --- a/app/src/main/java/com/example/studentfaceregistry/face/FaceAligner.kt +++ b/app/src/main/java/com/example/studentfaceregistry/face/FaceAligner.kt @@ -1,11 +1,10 @@ package com.example.studentfaceregistry.face import android.graphics.Bitmap +import android.graphics.Canvas import android.graphics.Matrix +import android.graphics.Paint import kotlin.math.absoluteValue -import kotlin.math.cos -import kotlin.math.sin -import kotlin.math.sqrt /** * 人脸对齐器 @@ -34,10 +33,18 @@ class FaceAligner { faceTop: Int ): Bitmap { // 计算眼睛在裁剪后图片中的相对位置 - val leftEyeRelX = leftEyeX - faceLeft - val leftEyeRelY = leftEyeY - faceTop - val rightEyeRelX = rightEyeX - faceLeft - val rightEyeRelY = rightEyeY - faceTop + var leftEyeRelX = leftEyeX - faceLeft + var leftEyeRelY = leftEyeY - faceTop + var rightEyeRelX = rightEyeX - faceLeft + var rightEyeRelY = rightEyeY - faceTop + if (leftEyeRelX > rightEyeRelX) { + val oldLeftX = leftEyeRelX + val oldLeftY = leftEyeRelY + leftEyeRelX = rightEyeRelX + leftEyeRelY = rightEyeRelY + rightEyeRelX = oldLeftX + rightEyeRelY = oldLeftY + } // 计算两眼连线的旋转角度（目标是让连线水平） val angle = calculateRotationAngle(leftEyeRelX, leftEyeRelY, rightEyeRelX, rightEyeRelY) @@ -58,6 +65,40 @@ class FaceAligner { return Bitmap.createBitmap(bitmap, 0, 0, bitmap.width, bitmap.height, matrix, true) } + fun alignFivePoint( + bitmap: Bitmap, + leftEyeX: Float, + leftEyeY: Float, + rightEyeX: Float, + rightEyeY: Float, + noseX: Float, + noseY: Float, + leftMouthX: Float, + leftMouthY: Float, + rightMouthX: Float, + rightMouthY: Float, + outputSize: Int = ARC_FACE_SIZE + ): Bitmap { + val (imageLeftEye, imageRightEye) = sortedByX(leftEyeX, leftEyeY, rightEyeX, rightEyeY) + val (imageLeftMouth, imageRightMouth) = sortedByX(leftMouthX, leftMouthY, rightMouthX, rightMouthY) + val src = floatArrayOf( + imageLeftEye.x, imageLeftEye.y, + imageRightEye.x, imageRightEye.y, + noseX, noseY, + imageLeftMouth.x, imageLeftMouth.y, + imageRightMouth.x, imageRightMouth.y + ) + val dst = arcFaceTemplate(outputSize) + val matrix = similarityTransform(src, dst) + val result = Bitmap.createBitmap(outputSize, outputSize, Bitmap.Config.ARGB_8888) + Canvas(result).drawBitmap( + bitmap, + matrix, + Paint(Paint.ANTI_ALIAS_FLAG or Paint.FILTER_BITMAP_FLAG) + ) + return result + } + /** * 计算两眼连线的旋转角度 */ @@ -67,6 +108,63 @@ class FaceAligner { return Math.atan2(dy.toDouble(), dx.toDouble()).toFloat() * 180f / Math.PI.toFloat() } + private fun sortedByX(x1: Float, y1: Float, x2: Float, y2: Float): Pair { + val first = Point(x1, y1) + val second = Point(x2, y2) + return if (x1 <= x2) first to second else second to first + } + + private fun arcFaceTemplate(outputSize: Int): FloatArray { + val scale = outputSize / ARC_FACE_SIZE.toFloat() + return FloatArray(ARC_FACE_TEMPLATE.size) { index -> ARC_FACE_TEMPLATE[index] * scale } + } + + private fun similarityTransform(src: FloatArray, dst: FloatArray): Matrix { + val count = src.size / 2 + var srcCenterX = 0f + var srcCenterY = 0f + var dstCenterX = 0f + var dstCenterY = 0f + for (index in 0 until count) { + srcCenterX += src[index * 2] + srcCenterY += src[index * 2 + 1] + dstCenterX += dst[index * 2] + dstCenterY += dst[index * 2 + 1] + } + srcCenterX /= count + srcCenterY /= count + dstCenterX /= count + dstCenterY /= count + + var numeratorA = 0f + var numeratorB = 0f + var denominator = 0f + for (index in 0 until count) { + val srcX = src[index * 2] - srcCenterX + val srcY = src[index * 2 + 1] - srcCenterY + val dstX = dst[index * 2] - dstCenterX + val dstY = dst[index * 2 + 1] - dstCenterY + numeratorA += srcX * dstX + srcY * dstY + numeratorB += srcX * dstY - srcY * dstX + denominator += srcX * srcX + srcY * srcY + } + + val a = numeratorA / denominator.coerceAtLeast(1e-6f) + val b = numeratorB / denominator.coerceAtLeast(1e-6f) + val translateX = dstCenterX - a * srcCenterX + b * srcCenterY + val translateY = dstCenterY - b * srcCenterX - a * srcCenterY + + return Matrix().apply { + setValues( + floatArrayOf( + a, -b, translateX, + b, a, translateY, + 0f, 0f, 1f + ) + ) + } + } + /** * 简单版本：仅根据估计的眼睛位置进行对齐 * 适用于没有启用关键点检测的场景 @@ -94,4 +192,17 @@ class FaceAligner { align(bitmap, leftEyeX, eyeY, rightEyeX, eyeY, faceLeft, faceTop) } } + + private data class Point(val x: Float, val y: Float) + + companion object { + private const val ARC_FACE_SIZE = 112 + private val ARC_FACE_TEMPLATE = floatArrayOf( + 38.2946f, 51.6963f, + 73.5318f, 51.5014f, + 56.0252f, 71.7366f, + 41.5493f, 92.3655f, + 70.7299f, 92.2041f + ) + } } diff --git a/app/src/main/java/com/example/studentfaceregistry/face/FaceEmbedder.kt b/app/src/main/java/com/example/studentfaceregistry/face/FaceEmbedder.kt index c57ee8f..63b2b5e 100644 --- a/app/src/main/java/com/example/studentfaceregistry/face/FaceEmbedder.kt +++ b/app/src/main/java/com/example/studentfaceregistry/face/FaceEmbedder.kt @@ -30,6 +30,7 @@ class FaceEmbedder(context: Context, private val modelType: ModelType = ModelTyp private val accelerator: Accelerator private val inputPixels: IntArray private val inputValues: FloatArray + private var debugEmbedCount = 0 // ArcFace 标准化参数 (RGB) private val mean = floatArrayOf(127.5f, 127.5f, 127.5f) @@ -85,12 +86,57 @@ class FaceEmbedder(context: Context, private val modelType: ModelType = ModelTyp } } + debugEmbedCount++ + val logInput = debugEmbedCount <= 30 || debugEmbedCount % 100 == 0 + if (logInput) { + var sum = 0f + var sumSq = 0f + for (value in inputValues) { + sum += value + sumSq += value * value + } + val mean = sum / inputValues.size + val variance = (sumSq / inputValues.size - mean * mean).coerceAtLeast(0f) + Log.i( + TAG, + String.format( + java.util.Locale.US, + "embed debug #%d 输入均值=%.3f 输入标准差=%.3f", + debugEmbedCount, + mean, + sqrt(variance) + ) + ) + } + // 执行推理 inputBuffer.writeFloat(inputValues) model.run(listOf(inputBuffer), listOf(outputBuffer)) + val rawOutput = outputBuffer.readFloat() + if (logInput) { + var sum = 0f + var sumSq = 0f + for (value in rawOutput) { + sum += value + sumSq += value * value + } + val mean = sum / rawOutput.size + val variance = (sumSq / rawOutput.size - mean * mean).coerceAtLeast(0f) + Log.i( + TAG, + String.format( + java.util.Locale.US, + "embed debug #%d 输出均值=%.3f 输出标准差=%.3f", + debugEmbedCount, + mean, + sqrt(variance) + ) + ) + } + // L2 归一化输出特征 - return l2Normalize(outputBuffer.readFloat()) + return l2Normalize(rawOutput) } fun warmUp() { @@ -138,19 +184,10 @@ class FaceEmbedder(context: Context, private val modelType: ModelType = ModelTyp } private fun createModel(context: Context): Pair { - return runCatching { - CompiledModel.create( - context.assets, - MODEL_ASSET_NAME, - CompiledModel.Options(setOf(Accelerator.GPU)) - ) to Accelerator.GPU - }.getOrElse { gpuError -> - Log.w(TAG, "GPU initialization failed, falling back to CPU", gpuError) - CompiledModel.create( - context.assets, - MODEL_ASSET_NAME, - CompiledModel.Options(setOf(Accelerator.CPU)) - ) to Accelerator.CPU - } + return CompiledModel.create( + context.assets, + MODEL_ASSET_NAME, + CompiledModel.Options(setOf(Accelerator.CPU)) + ) to Accelerator.CPU } } diff --git a/app/src/main/java/com/example/studentfaceregistry/face/FaceProcessor.kt b/app/src/main/java/com/example/studentfaceregistry/face/FaceProcessor.kt index d40400c..58ee6c2 100644 --- a/app/src/main/java/com/example/studentfaceregistry/face/FaceProcessor.kt +++ b/app/src/main/java/com/example/studentfaceregistry/face/FaceProcessor.kt @@ -57,17 +57,19 @@ class FaceProcessor(context: Context) : AutoCloseable { box.width() * box.height() } - // 裁剪人脸 - val croppedFace = Bitmap.createBitmap( - bitmap, - faceBox.left, - faceBox.top, - faceBox.width(), - faceBox.height() - ) + val alignedFace = alignFaceFromOriginal(bitmap, largestFace) ?: run { + // 裁剪人脸 + val croppedFace = Bitmap.createBitmap( + bitmap, + faceBox.left, + faceBox.top, + faceBox.width(), + faceBox.height() + ) - // 使用关键点进行对齐 - val alignedFace = alignFaceWithLandmarks(croppedFace, largestFace, faceBox.left, faceBox.top) + // 使用关键点进行对齐 + alignFaceWithLandmarks(croppedFace, largestFace, faceBox.left, faceBox.top) + } // 图像预处理 val preprocessedFace = preprocessor.preprocess(alignedFace) @@ -127,6 +129,12 @@ class FaceProcessor(context: Context) : AutoCloseable { * 提取指定人脸的特征 */ fun embedFace(bitmap: Bitmap, face: Face): FloatArray { + val alignedFromLandmarks = alignFaceFromOriginal(bitmap, face) + if (alignedFromLandmarks != null) { + val preprocessed = preprocessor.preprocessRealtime(alignedFromLandmarks) + return embedder.embed(preprocessed) + } + val crop = requireNotNull(safeCropRect(bitmap, face.boundingBox)) { "人脸区域超出图片边界。" } val cropped = Bitmap.createBitmap(bitmap, crop.left, crop.top, crop.width(), crop.height()) val aligned = alignFaceForRecognition(cropped, face, crop.left, crop.top) @@ -136,6 +144,9 @@ class FaceProcessor(context: Context) : AutoCloseable { fun embedFaceCandidates(bitmap: Bitmap, face: Face): List { val embeddings = mutableListOf() + alignFaceFromOriginal(bitmap, face)?.let { aligned -> + embeddings.add(embedder.embed(preprocessor.preprocessRealtime(aligned))) + } for (paddingRatio in FACE_PADDING_RATIOS) { val crop = safeCropRect(bitmap, face.boundingBox, paddingRatio) ?: continue val cropped = Bitmap.createBitmap(bitmap, crop.left, crop.top, crop.width(), crop.height()) @@ -146,6 +157,28 @@ class FaceProcessor(context: Context) : AutoCloseable { return embeddings } + private fun alignFaceFromOriginal(bitmap: Bitmap, face: Face): Bitmap? { + val leftEye = face.getLandmark(FaceLandmark.LEFT_EYE) ?: return null + val rightEye = face.getLandmark(FaceLandmark.RIGHT_EYE) ?: return null + val nose = face.getLandmark(FaceLandmark.NOSE_BASE) ?: return null + val leftMouth = face.getLandmark(FaceLandmark.MOUTH_LEFT) ?: return null + val rightMouth = face.getLandmark(FaceLandmark.MOUTH_RIGHT) ?: return null + + return aligner.alignFivePoint( + bitmap = bitmap, + leftEyeX = leftEye.position.x, + leftEyeY = leftEye.position.y, + rightEyeX = rightEye.position.x, + rightEyeY = rightEye.position.y, + noseX = nose.position.x, + noseY = nose.position.y, + leftMouthX = leftMouth.position.x, + leftMouthY = leftMouth.position.y, + rightMouthX = rightMouth.position.x, + rightMouthY = rightMouth.position.y + ) + } + fun warmUp() { embedder.warmUp() } diff --git a/app/src/main/java/com/example/studentfaceregistry/face/SmartEnrollment.kt b/app/src/main/java/com/example/studentfaceregistry/face/SmartEnrollment.kt index 88e29ed..5381135 100644 --- a/app/src/main/java/com/example/studentfaceregistry/face/SmartEnrollment.kt +++ b/app/src/main/java/com/example/studentfaceregistry/face/SmartEnrollment.kt @@ -101,20 +101,23 @@ class SmartEnrollment(context: Context) : AutoCloseable { return EnrollmentResult.Failure("照片模糊，请换更清晰的照片。") } - val embeddings = mutableListOf() - for (paddingRatio in IMAGE_PADDING_RATIOS) { - val cropRect = safeCropRect(bitmap, largestFace.face.boundingBox, paddingRatio) ?: continue - val croppedFace = cropFace(bitmap, cropRect) - val alignedFace = alignFace(croppedFace, largestFace.face, cropRect) - val preprocessedFace = preprocessor.preprocess(alignedFace) - embeddings.add(embedder.embed(preprocessedFace)) + val alignedFromLandmarks = alignFaceFromOriginal(bitmap, largestFace.face) + val cropEmbeddings = cropEmbeddingCandidates(bitmap, largestFace.face) + if (alignedFromLandmarks != null) { + val preprocessedFace = preprocessor.preprocess(alignedFromLandmarks) + val embedding = embedder.embed(preprocessedFace) + return EnrollmentResult.Success( + embedding, + EnrollmentSourceType.IMAGE, + extraEmbeddings = cropEmbeddings + ) } - if (embeddings.isEmpty()) { + if (cropEmbeddings.isEmpty()) { return EnrollmentResult.Failure("未检测到可用人脸，请换更清晰或更正面的照片。") } - val embedding = fuseEmbeddings(embeddings) + val embedding = fuseEmbeddings(cropEmbeddings) EnrollmentResult.Success(embedding, EnrollmentSourceType.IMAGE) } catch (e: Exception) { @@ -355,12 +358,30 @@ class SmartEnrollment(context: Context) : AutoCloseable { * 对单帧提取特征（裁剪 + 对齐 + 预处理 + 推理） */ private fun embedFrame(frameInfo: FaceFrameInfo): FloatArray { + val alignedFromLandmarks = alignFaceFromOriginal(frameInfo.bitmap, frameInfo.face) + if (alignedFromLandmarks != null) { + val preprocessed = preprocessor.preprocess(alignedFromLandmarks) + return embedder.embed(preprocessed) + } + val cropped = cropFace(frameInfo.bitmap, frameInfo.cropRect) val aligned = alignFace(cropped, frameInfo.face, frameInfo.cropRect) val preprocessed = preprocessor.preprocess(aligned) return embedder.embed(preprocessed) } + private fun cropEmbeddingCandidates(bitmap: Bitmap, face: Face): List { + return buildList { + for (paddingRatio in IMAGE_PADDING_RATIOS) { + val cropRect = safeCropRect(bitmap, face.boundingBox, paddingRatio) ?: continue + val croppedFace = cropFace(bitmap, cropRect) + val alignedFace = alignFace(croppedFace, face, cropRect) + val preprocessedFace = preprocessor.preprocess(alignedFace) + add(embedder.embed(preprocessedFace)) + } + } + } + /** * 按偏航角将帧特征分组：正脸（主特征）、左右侧脸（补充特征） */ @@ -471,6 +492,28 @@ class SmartEnrollment(context: Context) : AutoCloseable { ) } + private fun alignFaceFromOriginal(bitmap: Bitmap, face: Face): Bitmap? { + val leftEye = face.getLandmark(FaceLandmark.LEFT_EYE) ?: return null + val rightEye = face.getLandmark(FaceLandmark.RIGHT_EYE) ?: return null + val nose = face.getLandmark(FaceLandmark.NOSE_BASE) ?: return null + val leftMouth = face.getLandmark(FaceLandmark.MOUTH_LEFT) ?: return null + val rightMouth = face.getLandmark(FaceLandmark.MOUTH_RIGHT) ?: return null + + return aligner.alignFivePoint( + bitmap = bitmap, + leftEyeX = leftEye.position.x, + leftEyeY = leftEye.position.y, + rightEyeX = rightEye.position.x, + rightEyeY = rightEye.position.y, + noseX = nose.position.x, + noseY = nose.position.y, + leftMouthX = leftMouth.position.x, + leftMouthY = leftMouth.position.y, + rightMouthX = rightMouth.position.x, + rightMouthY = rightMouth.position.y + ) + } + /** * 解码 Bitmap */ @@ -522,7 +565,7 @@ class SmartEnrollment(context: Context) : AutoCloseable { private val IMAGE_PADDING_RATIOS = floatArrayOf(0.18f, 0.25f, 0.34f) private const val MIN_ENROLL_FACE_PX = 96 private const val MAX_ENROLL_YAW = 30f - private const val MAX_ENROLL_PITCH = 15f + private const val MAX_ENROLL_PITCH = 25f private const val BLUR_VARIANCE_THRESHOLD = 40.0 } } diff --git a/app/src/main/java/com/example/studentfaceregistry/upload/UploadServer.kt b/app/src/main/java/com/example/studentfaceregistry/upload/UploadServer.kt index 0f00530..d09b140 100644 --- a/app/src/main/java/com/example/studentfaceregistry/upload/UploadServer.kt +++ b/app/src/main/java/com/example/studentfaceregistry/upload/UploadServer.kt @@ -2,10 +2,12 @@ package com.example.studentfaceregistry.upload import android.content.Context import android.util.Base64 +import android.util.Log import com.example.studentfaceregistry.data.Student import com.example.studentfaceregistry.data.StudentRepository import com.example.studentfaceregistry.face.SmartEnrollment import kotlinx.coroutines.runBlocking +import kotlinx.coroutines.withTimeoutOrNull import org.json.JSONObject import java.io.BufferedInputStream import java.io.ByteArrayOutputStream @@ -58,13 +60,18 @@ class UploadServer( private fun handle(socket: Socket) { thread(name = "upload-request", isDaemon = true) { socket.use { client -> - val input = BufferedInputStream(client.getInputStream()) - val output = client.getOutputStream() - val request = readHttpRequest(input) - when { - request.method == "GET" && request.path == "/" -> respondHtml(output) - request.method == "POST" && request.path == "/upload" -> handleUpload(request, output) - else -> respondText(output, 404, "Not found", "text/plain; charset=utf-8") + client.soTimeout = SOCKET_READ_TIMEOUT_MS + runCatching { + val input = BufferedInputStream(client.getInputStream()) + val output = client.getOutputStream() + val request = readHttpRequest(input) + when { + request.method == "GET" && request.path == "/" -> respondHtml(output) + request.method == "POST" && request.path == "/upload" -> handleUpload(request, output) + else -> respondText(output, 404, "Not found", "text/plain; charset=utf-8") + } + }.onFailure { + Log.w(TAG, "Upload request failed", it) } } } @@ -133,33 +140,78 @@ class UploadServer(