commit 22/05/2025

src/main/java/id/co/gtc/cctvwithcuda/MainController.java

@@ -7,6 +7,9 @@ import javafx.scene.control.TextField;
 import javafx.scene.image.Image;
 import javafx.scene.image.ImageView;
 import javafx.scene.layout.StackPane;
+import org.opencv.core.Rect;
+import org.opencv.core.Scalar;
+import org.opencv.imgproc.Imgproc;
 import org.opencv.osgi.OpenCVNativeLoader;
 
 import java.time.LocalDateTime;
@@ -106,17 +109,50 @@ public class MainController {
                 },
                 (frame) -> {
                     yoloDetector.Detect(frame, (detection)->{
-                        String dimension = "Dimension : "+detection.dims();
+
-                        String imagesize = "Image Size : "+detection.size(0);
+                        if (detection!=null && !detection.empty()){
-                        String imagechannels = "Image Channels : "+detection.size(1);
+                            for (int i=0; i<detection.rows();i++){
-                        String imageheight = "Image Height : "+detection.size(2);
+
-                        String imagewidth = "Image Width : "+detection.size(3);
+                                float[] row = new float[84];
-                        System.out.println("Detection : ");
+                                detection.get(i, 0, row);
-                        System.out.println(dimension);
+                                float confidence = row[4];
-                        System.out.println(imagesize);
+                                if (confidence<0.3) continue;
-                        System.out.println(imagechannels);
+                                int classId = -1;
-                        System.out.println(imageheight);
+                                float maxScore = -1;
-                        System.out.println(imagewidth);
+                                for (int j = 5; j < row.length; j++) {
+                                    float score = row[j];
+                                    if (score > maxScore) {
+                                        maxScore = score;
+                                        classId = j - 5;
+                                    }
+                                }
+
+                                System.out.println("MaxScore: " + maxScore + ", ClassId: " + classId);
+                                if (classId < 0) continue;
+                                if (classId >= 80) continue;
+                                String className = yoloDetector.getClassName(classId);
+                                if (className == null) continue;
+                                System.out.println("ClassName: " + className + ", Confidence: " + confidence);
+
+                                float cx = row[0] * frame.cols();
+                                float cy = row[1] * frame.rows();
+                                float w  = row[2] * frame.cols();
+                                float h  = row[3] * frame.rows();
+
+                                int x = (int)(cx - w / 2);
+                                int y  = (int)(cy - h / 2);
+
+                                System.out.println("x: " + x + ", y: " + y + ", w: " + w + ", h: " + h);
+
+
+                                Rect rect = new Rect(x, y, (int)w, (int)h);
+                                Scalar color = new Scalar(0, 255, 0);
+                                Imgproc.rectangle(frame, rect, color, 2);
+                            }
+                        }
+
+
+
                     });
                     Image img = SomeCodes.MatToImage(frame);
                     if (img!=null) {

src/main/java/id/co/gtc/cctvwithcuda/SomeCodes.java

@@ -6,16 +6,20 @@ import javafx.scene.control.Button;
 import javafx.scene.control.Label;
 import javafx.scene.image.Image;
 import javafx.scene.image.ImageView;
-import org.opencv.core.Mat;
+import org.opencv.core.*;
-import org.opencv.core.MatOfByte;
 import org.opencv.imgcodecs.Imgcodecs;
+import org.opencv.imgproc.Imgproc;
 
 import java.io.ByteArrayInputStream;
 import java.io.File;
+import java.io.IOException;
 import java.io.InputStream;
 import java.nio.file.Files;
+import java.nio.file.Paths;
 import java.time.LocalDateTime;
 import java.time.format.DateTimeFormatter;
+import java.util.ArrayList;
+import java.util.List;
 
 public class SomeCodes {
     public final static String currentDirectory = System.getProperty("user.dir");
@@ -98,4 +102,50 @@ public class SomeCodes {
         return null;
     }
 
+    public static Mat Convert_BGRA_To_BGR(Mat src){
+        if (src.channels() == 4){
+            Mat dst = new Mat();
+            Imgproc.cvtColor(src, dst, Imgproc.COLOR_BGRA2BGR);
+            return dst;
+        } else return src;
+
+    }
+
+    public static boolean IsSquared(Mat src){
+        if (src != null && !src.empty()){
+            int width = src.cols();
+            int height = src.rows();
+            return width == height;
+        }
+        return false;
+    }
+
+    public static Mat FormatToSquare(Mat src){
+        if (src != null && !src.empty()){
+            int width = src.cols();
+            int height = src.rows();
+            int size = Math.max(width, height);
+
+            int right = size - width;
+            int bottom = size - height;
+
+            Scalar paddingColor = new Scalar(0, 0, 0); // Black background
+
+            Mat dst = new Mat();
+            Core.copyMakeBorder(src, dst, 0, bottom, 0, right, Core.BORDER_CONSTANT, paddingColor);
+            return dst;
+        }
+        return null;
+    }
+
+    public static List<String> ReadLines(String filename){
+        try {
+            return Files.readAllLines(Paths.get(currentDirectory,filename));
+        } catch (IOException e) {
+            System.out.println("Error reading file: " + filename + ", Message: " + e.getMessage());
+        }
+        return new ArrayList<>();
+    }
+
+
 }

src/main/java/id/co/gtc/cctvwithcuda/YoloDetector.java

@@ -1,18 +1,21 @@
 package id.co.gtc.cctvwithcuda;
 
+import org.opencv.core.Core;
 import org.opencv.core.Mat;
 import org.opencv.core.Scalar;
 import org.opencv.core.Size;
 import org.opencv.dnn.Dnn;
 import org.opencv.dnn.Net;
 
+import java.nio.file.Files;
+import java.nio.file.Paths;
+import java.util.List;
 import java.util.function.Consumer;
 
 @SuppressWarnings("FieldCanBeLocal")
 public class YoloDetector {
 
-
+    private List<String> classNames = null;
-
     private Net net = null;
     private boolean loaded = false;
 
@@ -20,6 +23,14 @@ public class YoloDetector {
      * Create a YoloDetector object
      */
     public YoloDetector(String yolofile){
+        SomeCodes.ExtractResource("/coco.names");
+        classNames = SomeCodes.ReadLines("coco.names");
+        if (!classNames.isEmpty()){
+            System.out.println("YoloDetector class names loaded from coco.names, length: " + classNames.size());
+        } else {
+            System.out.println("YoloDetector class names not loaded from coco.names");
+        }
+
         if (yolofile!=null && !yolofile.isEmpty()){
             if (!yolofile.startsWith("/")) yolofile = "/" + yolofile;
             String yolofilepath = SomeCodes.ExtractResource(yolofile);
@@ -34,15 +45,26 @@ public class YoloDetector {
                 }
             }
         }
-        SomeCodes.ExtractResource("/coco.names");
+
         System.out.println("YoloDetector failed to load from " + yolofile);
     }
 
+    public String getClassName(int index){
+        if (classNames!=null && index>=0 && index<classNames.size()){
+            return classNames.get(index);
+        } else {
+            return null;
+        }
+    }
+
+
+
     private final double scale = 1/255.0;
     private final boolean swapRGB = true;
     private final boolean crop = false;
     private final Size blobSize = new Size(640, 640);
     private final Scalar blobScalar = new Scalar(0, 0, 0);
+    private final boolean transposed = true;
 
     /**
      * Detect objects in the given frame
@@ -51,12 +73,30 @@ public class YoloDetector {
      */
     public void Detect(Mat frame, Consumer<Mat> onDetect){
         if (loaded){
+            //System.out.println("frame width: " + frame.width() + ", height: " + frame.height()+" channels: " + frame.channels());
+            if (frame.channels()==4) frame = SomeCodes.Convert_BGRA_To_BGR(frame);
+            if (!SomeCodes.IsSquared(frame)) frame = SomeCodes.FormatToSquare(frame);
+            //System.out.println("new frame width: " + frame.width() + ", height: " + frame.height()+" channels: " + frame.channels());
             Mat blob = Dnn.blobFromImage(frame, scale, blobSize, blobScalar, swapRGB, crop);
             net.setInput(blob);
             Mat detections = net.forward();
-            if (detections!=null && !detections.empty()){
+            if (transposed){
+                // Fix shape from [1, 84, 8400] → [8400, 84]
+                Mat reshaped = detections.reshape(1, 84);  // [84 x 8400]
+                Mat transposed = new Mat();
+                Core.transpose(reshaped, transposed); // [8400 x 84]
+                if (!transposed.empty()){
+                    if (onDetect!=null) onDetect.accept(transposed);
+                }
+                reshaped.release();
+            } else {
                 if (onDetect!=null) onDetect.accept(detections);
             }
+
         }
     }
+
+
+
+
 }