From b001537103af607630dc28b69d005487064fbd67 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E2=80=9Ckamon?= <“kamon.sourabie@imt-atlantique.net”>
Date: Tue, 7 Jan 2025 01:22:02 +0100
Subject: [PATCH] amelioration de l'affichage des cadres et noms des especes
avec un leger bug dans le cas d'especes non reconnues
---
.idea/compiler.xml | 2 +-
.idea/deploymentTargetSelector.xml | 4 +-
.idea/misc.xml | 3 +-
.../example/myapplication/MainActivity.java | 39 ---
.../myapplication/TestModeleTflite.java | 225 ++++++++++--------
5 files changed, 132 insertions(+), 141 deletions(-)
diff --git a/.idea/compiler.xml b/.idea/compiler.xml
index b86273d..b589d56 100644
--- a/.idea/compiler.xml
+++ b/.idea/compiler.xml
@@ -1,6 +1,6 @@
<?xml version="1.0" encoding="UTF-8"?>
<project version="4">
<component name="CompilerConfiguration">
- <bytecodeTargetLevel target="21" />
+ <bytecodeTargetLevel target="17" />
</component>
</project>
\ No newline at end of file
diff --git a/.idea/deploymentTargetSelector.xml b/.idea/deploymentTargetSelector.xml
index bcc86d0..7762630 100644
--- a/.idea/deploymentTargetSelector.xml
+++ b/.idea/deploymentTargetSelector.xml
@@ -4,10 +4,10 @@
<selectionStates>
<SelectionState runConfigName="app">
<option name="selectionMode" value="DROPDOWN" />
- <DropdownSelection timestamp="2025-01-04T02:38:58.283128700Z">
+ <DropdownSelection timestamp="2025-01-07T00:07:51.508741600Z">
<Target type="DEFAULT_BOOT">
<handle>
- <DeviceId pluginId="PhysicalDevice" identifier="serial=R39M30F5B2T" />
+ <DeviceId pluginId="LocalEmulator" identifier="path=C:\Users\Prosp\.android\avd\Pixel_8a_API_35.avd" />
</handle>
</Target>
</DropdownSelection>
diff --git a/.idea/misc.xml b/.idea/misc.xml
index b2c751a..0ad17cb 100644
--- a/.idea/misc.xml
+++ b/.idea/misc.xml
@@ -1,6 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
<project version="4">
<component name="ExternalStorageConfigurationManager" enabled="true" />
- <component name="ProjectRootManager" version="2" languageLevel="JDK_21" default="true" project-jdk-name="jbr-21" project-jdk-type="JavaSDK">
+ <component name="ProjectRootManager" version="2" languageLevel="JDK_17" default="true" project-jdk-name="jbr-17" project-jdk-type="JavaSDK">
<output url="file://$PROJECT_DIR$/build/classes" />
</component>
<component name="ProjectType">
diff --git a/app/src/main/java/com/example/myapplication/MainActivity.java b/app/src/main/java/com/example/myapplication/MainActivity.java
index 12f9d38..2393421 100644
--- a/app/src/main/java/com/example/myapplication/MainActivity.java
+++ b/app/src/main/java/com/example/myapplication/MainActivity.java
@@ -30,7 +30,6 @@ public class MainActivity extends AppCompatActivity {
setContentView(R.layout.activity_main);
-
int delayMillis = 3000;
new Handler().postDelayed(new Runnable() {
@@ -46,44 +45,6 @@ public class MainActivity extends AppCompatActivity {
}
- public void testPredict(View v){
- TextView textView = findViewById(R.id.testInference);
- TestModeleTflite testModeleTflite = null;
- try {
- testModeleTflite = new TestModeleTflite(this);
- String currentDir = System.getProperty("user.dir");
- String imagePath = "mouette_rieuse.jpg";
- Bitmap inputImage = TestModeleTflite.loadImage(imagePath, this);
-
- if (inputImage == null) {
- System.out.println("Failed to load the image. Check the file path.");
- return;
- }
-
- // Perform prediction
- TensorBuffer outputBuffer = testModeleTflite.predict(inputImage);
-
- // Process the output to get bounding boxes
- List<BoundingBox> boxes = testModeleTflite.processOutput(outputBuffer);
-
- if (boxes == null || boxes.isEmpty()) {
- textView.setText("No objects detected in the image.");
- } else {
- // Output results to the console
- for (BoundingBox box : boxes) {
- System.out.println("Class: " + box.clsName);
- System.out.println("Confidence: " + box.cnf);
- System.out.println("----------------------------");
- textView.setText("Class: " + box.clsName + " Confidence : " + box.cnf);
- }
-
- }
- } catch (IOException e) {
- throw new RuntimeException(e);
- }
-
- //textView.setText("bonjour");
- }
}
diff --git a/app/src/main/java/com/example/myapplication/TestModeleTflite.java b/app/src/main/java/com/example/myapplication/TestModeleTflite.java
index 1cd882e..6b10aca 100644
--- a/app/src/main/java/com/example/myapplication/TestModeleTflite.java
+++ b/app/src/main/java/com/example/myapplication/TestModeleTflite.java
@@ -4,7 +4,6 @@ import static android.graphics.Paint.*;
import android.content.Context;
import android.graphics.Bitmap;
-import android.graphics.BitmapFactory;
import android.graphics.Canvas;
import android.graphics.Color;
import android.graphics.Paint;
@@ -20,8 +19,6 @@ import org.tensorflow.lite.support.image.TensorImage;
import org.tensorflow.lite.support.tensorbuffer.TensorBuffer;
import java.io.BufferedReader;
-import java.io.File;
-import java.io.FileInputStream;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
@@ -31,6 +28,10 @@ import java.util.HashMap;
import java.util.List;
import java.util.Map;
+/**
+ * A class for performing species recognition using a TensorFlow Lite model.
+ * Handles model loading, inference, output processing, and annotation of input images.
+ */
public class TestModeleTflite {
private static final String MODEL_PATH = "model_float32.tflite";
private static final String LABEL_PATH = "labels.txt";
@@ -50,16 +51,22 @@ public class TestModeleTflite {
private List<String> labels = new ArrayList<>();
private ImageProcessor imageProcessor;
+ /**
+ * Initializes the TensorFlow Lite model interpreter and loads associated resources.
+ *
+ * @param context The application context for accessing assets.
+ * @throws IOException If the model or label files cannot be loaded.
+ */
public TestModeleTflite(Context context) throws IOException {
- // Load the model
+ // Load the TensorFlow Lite model
MappedByteBuffer model = FileUtil.loadMappedFile(context, MODEL_PATH);
- // Set interpreter options
+ // Configure the interpreter with multithreading options
Interpreter.Options options = new Interpreter.Options();
options.setNumThreads(4);
interpreter = new Interpreter(model, options);
- // Get tensor shapes
+ // Extract tensor shapes from the model
int[] inputShape = interpreter.getInputTensor(0).shape();
int[] outputShape = interpreter.getOutputTensor(0).shape();
@@ -68,7 +75,7 @@ public class TestModeleTflite {
numChannel = outputShape[1];
numElements = outputShape[2];
- // Load labels
+ // Load the label list
try (InputStream inputStream = context.getAssets().open(LABEL_PATH);
BufferedReader reader = new BufferedReader(new InputStreamReader(inputStream))) {
String line;
@@ -77,54 +84,95 @@ public class TestModeleTflite {
}
}
- // Initialize image processor
+ // Configure preprocessing steps for the input image
imageProcessor = new ImageProcessor.Builder()
.add(new NormalizeOp(INPUT_MEAN, INPUT_STANDARD_DEVIATION))
.add(new CastOp(INPUT_IMAGE_TYPE))
.build();
}
+ /**
+ * Prepares an input image and performs inference using the TensorFlow Lite model.
+ *
+ * @param bitmap The input image to analyze.
+ * @return A TensorBuffer containing the model's raw output.
+ */
public TensorBuffer predict(Bitmap bitmap) {
- // Resize bitmap to model's input size
Bitmap resizedBitmap = Bitmap.createScaledBitmap(bitmap, tensorWidth, tensorHeight, false);
- // Preprocess input image
TensorImage tensorImage = new TensorImage(DataType.FLOAT32);
tensorImage.load(resizedBitmap);
TensorImage processedImage = imageProcessor.process(tensorImage);
- // Prepare output buffer
TensorBuffer outputBuffer = TensorBuffer.createFixedSize(
new int[]{1, numChannel, numElements},
OUTPUT_IMAGE_TYPE
);
- // Run inference
interpreter.run(processedImage.getBuffer(), outputBuffer.getBuffer());
-
return outputBuffer;
}
- public List<String> getLabels() {
- return labels;
- }
-
- public void close() {
- if (interpreter != null) {
- interpreter.close();
- interpreter = null;
- }
- }
-
+ /**
+ * Processes the raw model output to generate bounding boxes for detected objects.
+ *
+ * @param outputBuffer The TensorBuffer containing raw model output.
+ * @return A list of bounding boxes for detected objects.
+ */
public List<BoundingBox> processOutput(TensorBuffer outputBuffer) {
float[] outputArray = outputBuffer.getFloatArray();
return bestBox(outputArray);
}
+ /**
+ * Annotates an input image with bounding boxes for detected objects.
+ *
+ * @param inputImage The original image to annotate.
+ * @param boxes The bounding boxes of detected objects.
+ * @return A new image with bounding boxes drawn.
+ */
public Bitmap annotateImage(Bitmap inputImage, List<BoundingBox> boxes) {
return drawBoundingBoxes(inputImage, boxes);
}
+ /**
+ * Groups detections by class and constructs a structured output image and metadata.
+ *
+ * @param inputImage The input image for recognition.
+ * @return An OutputPredictedImage containing the annotated image and prediction results.
+ * @throws NoSpeciesRecognizedException If no objects are detected.
+ */
+ public OutputPredictedImage recognizeSpeciesClass(Bitmap inputImage) throws NoSpeciesRecognizedException {
+ Map<String, RecognizeSpecie> predictionResults = new HashMap<>();
+ TensorBuffer outputBuffer = predict(inputImage);
+ List<BoundingBox> boxes = processOutput(outputBuffer);
+
+ if (boxes != null) {
+ System.out.println("not empty");
+ for (BoundingBox box : boxes) {
+ predictionResults.computeIfAbsent(box.clsName, k -> new RecognizeSpecie(k, new ArrayList<>()))
+ .getSpecieBoxes().add(box);
+ }
+ Bitmap annotatedImage = annotateImage(inputImage, boxes);
+ return new OutputPredictedImage(annotatedImage, predictionResults);
+ } else {
+ throw new NoSpeciesRecognizedException();
+ }
+ }
+
+ /**
+ * Releases the resources held by the TensorFlow Lite interpreter.
+ */
+ public void close() {
+ if (interpreter != null) {
+ interpreter.close();
+ interpreter = null;
+ }
+ }
+
+ /**
+ * Filters bounding boxes based on confidence and non-maximum suppression (NMS).
+ */
private List<BoundingBox> bestBox(float[] array) {
List<BoundingBox> boundingBoxes = new ArrayList<>();
@@ -161,26 +209,29 @@ public class TestModeleTflite {
}
}
- if (boundingBoxes.isEmpty()) return null;
-
return applyNMS(boundingBoxes);
}
+ /**
+ * Applies non-maximum suppression (NMS) to eliminate overlapping boxes.
+ */
private List<BoundingBox> applyNMS(List<BoundingBox> boxes) {
List<BoundingBox> sortedBoxes = new ArrayList<>(boxes);
- sortedBoxes.sort((b1, b2) -> Float.compare(b2.cnf, b1.cnf)); // Sort by confidence descending
+ sortedBoxes.sort((b1, b2) -> Float.compare(b2.cnf, b1.cnf));
List<BoundingBox> selectedBoxes = new ArrayList<>();
while (!sortedBoxes.isEmpty()) {
BoundingBox first = sortedBoxes.remove(0);
selectedBoxes.add(first);
-
sortedBoxes.removeIf(nextBox -> calculateIoU(first, nextBox) >= IOU_THRESHOLD);
}
return selectedBoxes;
}
+ /**
+ * Calculates the Intersection over Union (IoU) between two bounding boxes.
+ */
private float calculateIoU(BoundingBox box1, BoundingBox box2) {
float x1 = Math.max(box1.x1, box2.x1);
float y1 = Math.max(box1.y1, box2.y1);
@@ -194,8 +245,10 @@ public class TestModeleTflite {
return intersectionArea / (box1Area + box2Area - intersectionArea);
}
-
-
+ /**
+ * Draws bounding boxes on an image.
+ */
+ /*
private Bitmap drawBoundingBoxes(Bitmap bitmap, List<BoundingBox> boxes) {
Bitmap mutableBitmap = bitmap.copy(Bitmap.Config.ARGB_8888, true);
Canvas canvas = new Canvas(mutableBitmap);
@@ -204,7 +257,7 @@ public class TestModeleTflite {
paint.setColor(Color.RED);
paint.setStyle(Style.STROKE);
paint.setStrokeWidth(8f);
-
+ /*
Paint textPaint = new Paint();
textPaint.setColor(Color.WHITE);
textPaint.setTextSize(40f);
@@ -218,89 +271,65 @@ public class TestModeleTflite {
box.y2 * mutableBitmap.getHeight()
);
canvas.drawRect(rect, paint);
- canvas.drawText(box.clsName, rect.left, rect.bottom, textPaint);
- //canvas.drawText(new StringBuilder().append("prob : ").append(String.format("%.2f", box.cnf)).toString(), rect.left, rect.top, textPaint);
+ //canvas.drawText(box.clsName, rect.left, rect.bottom, textPaint);
}
return mutableBitmap;
- }
+ }*/
- /**
- * Loads an image from the specified file path and converts it to a Bitmap.
- *
- * @param fileName Path to the image file.
- * @return Bitmap representation of the image or null if an error occurs.
- */
- public static Bitmap loadImage(String fileName, Context context) {
- try (InputStream inputStream = context.getAssets().open(fileName)) {
- return BitmapFactory.decodeStream(inputStream);
- } catch (IOException e) {
- e.printStackTrace();
- return null;
- }
- }
+ private Bitmap drawBoundingBoxes(Bitmap bitmap, List<BoundingBox> boxes) {
+ Bitmap mutableBitmap = bitmap.copy(Bitmap.Config.ARGB_8888, true);
+ Canvas canvas = new Canvas(mutableBitmap);
- public Bitmap recognizeSpecies(Bitmap inputImage){
- TensorBuffer outputBuffer = predict(inputImage);
- List<BoundingBox> boxes = processOutput(outputBuffer);
- return annotateImage(inputImage, boxes);
- }
+ // Scale stroke width dynamically based on image resolution
+ float strokeWidth = Math.max(mutableBitmap.getWidth(), mutableBitmap.getHeight()) / 470.0f;
- public OutputPredictedImage recognizeSpeciesClass(Bitmap inputImage) throws NoSpeciesRecognizedException {
- //ArrayList<String> speciesNames = new ArrayList<>();
- List<RecognizeSpecie> specieBoxes= new ArrayList<RecognizeSpecie>();
- Map<String, RecognizeSpecie> predictionResults = new HashMap<>();
+ Paint paint = new Paint();
+ paint.setColor(Color.RED);
+ paint.setStyle(Style.STROKE);
+ paint.setStrokeWidth(strokeWidth); // Dynamically set stroke width
- TensorBuffer outputBuffer = predict(inputImage);
- List<BoundingBox> boxes = processOutput(outputBuffer);
- if(boxes != null){
- for(BoundingBox box:boxes){
- if(predictionResults.containsKey(box.clsName)){
- predictionResults.get(box.clsName).getSpecieBoxes().add(box);
- }else{
- predictionResults.put(box.clsName, new RecognizeSpecie(box.clsName, new ArrayList<BoundingBox>()));
- predictionResults.get(box.clsName).getSpecieBoxes().add(box);
- }
+ Paint textPaint = new Paint();
+ textPaint.setColor(Color.WHITE);
+ textPaint.setStyle(Style.FILL);
- }
- Bitmap annotatedImage = annotateImage(inputImage, boxes);
- return new OutputPredictedImage(annotatedImage, predictionResults);
- }else throw new NoSpeciesRecognizedException();
- }
+ for (BoundingBox box : boxes) {
+ RectF rect = new RectF(
+ box.x1 * mutableBitmap.getWidth(),
+ box.y1 * mutableBitmap.getHeight(),
+ box.x2 * mutableBitmap.getWidth(),
+ box.y2 * mutableBitmap.getHeight()
+ );
+ // Draw the bounding box
+ canvas.drawRect(rect, paint);
- public static void main(String [] args){
- /*
- TestModeleTflite testModeleTflite = new TestModeleTflite(this);
- String imagePath = "../../res/drawable/bernache_cravant.jpg";
- Bitmap inputImage = TestModeleTflite.loadImage(imagePath);
+ // Dynamically adjust text size to fit within the box
+ float boxWidth = rect.width();
+ float boxHeight = rect.height();
+ float maxTextSize = boxHeight * 0.2f; // Limit text size to a fraction of the box height
+ textPaint.setTextSize(maxTextSize);
- if (inputImage == null) {
- System.out.println("Failed to load the image. Check the file path.");
- return;
+ // Measure text width and adjust if necessary
+ float textWidth = textPaint.measureText(box.clsName);
+ if (textWidth > boxWidth) {
+ textPaint.setTextSize(maxTextSize * (boxWidth / textWidth));
+ }
+
+ // Draw the text at the bottom of the box
+ canvas.drawText(box.clsName, rect.left, rect.bottom - 5, textPaint); // Adjust -5 for padding
}
- // Perform prediction
- TensorBuffer outputBuffer = testModeleTflite.predict(inputImage);
+ return mutableBitmap;
+ }
- // Process the output to get bounding boxes
- List<BoundingBox> boxes = testModeleTflite.processOutput(outputBuffer);
- if (boxes == null || boxes.isEmpty()) {
- System.out.println("No objects detected in the image.");
- } else {
- // Output results to the console
- for (BoundingBox box : boxes) {
- System.out.println("Class: " + box.clsName);
- System.out.println("Confidence: " + box.cnf);
- System.out.println("----------------------------");
- }
- }*/
+
+ /**
+ * Main method for debugging or testing purposes.
+ */
+ public static void main(String[] args) {
String currentDir = System.getProperty("user.dir");
System.out.println("Current working directory: " + currentDir);
}
-
-
-
}
-
--
GitLab