图像识别 无人机巡逻罂粟智能识别系统 springboot 框架 vue

package com.drone.poppy.detection.service.impl; import com.drone.poppy.detection.service.ImageRecognitionService; import org.springframework.stereotype.Service; import org.springframework.web.multipart.MultipartFile; import org.tensorflow.Graph; import org.tensorflow.Session; import org.tensorflow.Tensor; import org.tensorflow.TensorFlow; import java.io.IOException; import java.nio.file.Files; import java.nio.file.Path; import java.nio.file.Paths; import java.util.HashMap; import java.util.Map; @Service public class TensorFlowImageRecognitionServiceImpl implements ImageRecognitionService { private static final String MODEL_PATH = "model/poppy_detection_model.pb"; private static final String LABEL_PATH = "model/labels.txt"; private static final int INPUT_SIZE = 224; private static final int CHANNELS = 3; private byte[] graphDef; private String[] labels; public TensorFlowImageRecognitionServiceImpl() { try { loadModel(); loadLabels(); } catch (IOException e) { e.printStackTrace(); } } private void loadModel() throws IOException { Path modelPath = Paths.get(MODEL_PATH); this.graphDef = Files.readAllBytes(modelPath); } private void loadLabels() throws IOException { Path labelsPath = Paths.get(LABEL_PATH); this.labels = new String(Files.readAllBytes(labelsPath)).split("\n"); } @Override public Map<String, Object> recognizePoppy(MultipartFile imageFile) throws IOException { Path tempFile = Files.createTempFile("poppy-image-", ".jpg"); imageFile.transferTo(tempFile.toFile()); return recognizePoppyByPath(tempFile.toString()); } @Override public Map<String, Object> recognizePoppyByPath(String imagePath) throws IOException { Map<String, Object> result = new HashMap<>(); try (Graph g = new Graph()) { g.importGraphDef(graphDef); try (Session s = new Session(g); Tensor<Float> inputTensor = constructAndExecuteGraphToNormalizeImage(imagePath)) { Tensor<Float> output = s.runner() .feed("input", inputTensor) .fetch("output") .run() .get(0) .expect(Float.class); final long[] rshape = output.shape(); if (output.numDimensions() != 2 || rshape[0] != 1) { throw new RuntimeException( String.format("Expected model to produce a [1 N] shaped tensor where N is the number of labels, instead it produced one with shape %s", java.util.Arrays.toString(rshape))); } int nlabels = (int) rshape[1]; float[] probabilities = output.copyTo(new float[1][nlabels])[0]; int maxIndex = 0; float maxProbability = probabilities[0]; for (int i = 1; i < probabilities.length; i++) { if (probabilities[i] > maxProbability) { maxProbability = probabilities[i]; maxIndex = i; } } result.put("hasPoppy", labels[maxIndex].equals("poppy")); result.put("probability", maxProbability); result.put("label", labels[maxIndex]); } } return result; } private Tensor<Float> constructAndExecuteGraphToNormalizeImage(String imagePath) throws IOException { try (Graph g = new Graph()) { GraphBuilder b = new GraphBuilder(g); // 定义图操作：读取、解码和缩放图像 final String inputName = "file_reader"; b.readFile(inputName, imagePath); String outputName = "normalized"; b.normalizeImage(inputName, outputName, INPUT_SIZE, INPUT_SIZE, CHANNELS); try (Session s = new Session(g)) { return s.runner().fetch(outputName).run().get(0).expect(Float.class); } } } // 用于构建TensorFlow图的辅助类 static class GraphBuilder { private final Graph g; GraphBuilder(Graph g) { this.g = g; } Tensor<String> constant(String name, String value) { try (Tensor<String> t = Tensor.create(value, String.class)) { return g.opBuilder("Const", name) .setAttr("dtype", t.dataType()) .setAttr("value", t) .build() .output(0); } } Tensor<Float> normalizeImage(String inputName, String outputName, long width, long height, long channels) { Tensor<String> fileReader = constant("file_reader", inputName); Tensor<Float> decodedImage = decodeImage(fileReader); Tensor<Float> resizedImage = resizeImage(decodedImage, width, height); return g.opBuilder("Cast", outputName) .addInput(resizedImage) .setAttr("DstT", org.tensorflow.DataType.FLOAT) .build() .output(0); } Tensor<Float> decodeImage(Tensor<String> fileReader) { Tensor<Float> decodedImage = g.opBuilder("DecodeJpeg", "decode_jpeg") .addInput(fileReader) .setAttr("channels", 3L) .build() .output(0); return g.opBuilder("Cast", "float_casted") .addInput(decodedImage) .setAttr("DstT", org.tensorflow.DataType.FLOAT) .build() .output(0); } Tensor<Float> resizeImage(Tensor<Float> decodedImage, long width, long height) { Tensor<Long> desiredSize = g.opBuilder("Const", "size") .setAttr("dtype", org.tensorflow.DataType.INT32) .setAttr("value", Tensor.create(new int[]{(int) width, (int) height}, new long[]{2})) .build() .output(0); return g.opBuilder("ResizeBilinear", "resize") .addInput(decodedImage) .addInput(desiredSize) .build() .output(0); } Tensor<String> readFile(String name, String path) { return g.opBuilder("ReadFile", name) .addInput(constant(name + "_path", path)) .build() .output(0); } } }