Real time object detection Android application using OpenCV 4.1 and YOLO.

1. Install Android Studio on your computer.

2. Prepare Android device to run application.

3. Dowload OpenCV SDK from here

4. Open Android Studio and import this project & build.

5. From AndroidStudio top-menù select New -> Import Module and select your path to OpenCV sdk folder (i.e /where_opencv_saved/OpenCV-android-sdk/sdk) and rename module as OpenCV.

6. After load OpenCV module, re-build project.

1. From AndroidStudio top-menù select File -> Project Structure

2. Navigate to Dependencies and click on app. On the right panel there's a plus button + for add Dependency. Click on it and choose Module Dependecy.

3. Select OpenCV module loaded before.

4. Click Ok and Apply changes & Build project.

Load convolutional net from *.cfg and *.weights files and read labels name (COCO Dataset) 
in assets folder when calls onCameraViewStarted() using Dnn.readNetFromDarknet(String path_cfg, String path_weights).
NOTE: this repo doesn't contain weights file. You have to download it from YOLO site.

Iteratively generate a frame from CameraBridgeViewBase preview and analize it as an image. 
Real time detection and the frames flow generation is managed by onCameraFrame(CvCameraViewFrame inputFrame). 
Preview frame is translate in a Mat matrix and set as input for Dnn.blobFromImage(frame, scaleFactor, frame_size, mean, true, false) to preprocess frames. 
Note that frame_size is 416x416 for YOLO Model (you can find input dimension in *.cfg file). 
We can change the size by adding or subtracting by a factor of 32. 
Reducing the framesize increases the performance but worsens the accuracy. 
The detection phase is implemented by net.forward(List<Mat> results, List<String> outNames) that runs forward pass to compute output of layer with name outName. 
In results the method writes all detections in preview frame as Mat objects. 
Theese Mat instances contain all information such as positions and labels of detected objects.

Performing Non Maximum Suppression by YOLO, 
in List<Mat> results are stored all coordinates of optimal bounding boxes 
(the first 4 numbers are [center_x, center_y, width, height], 
followed by all class probabilities). 
classId is the corresponding index for label of detection in COCO Dataset list className.

<uses-permission android:name="android.permission.INTERNET"></uses-permission>

void connect(){
        mHandler = new Handler(Looper.getMainLooper());
        Log.w("connect","서버연결");
        // 받아오는거
        Thread checkUpdate = new Thread() {
            public void run() {
                // 서버 접속
                String newip = "210.102.181.248";
                int port = 7002;
                try {
                    socket = new Socket(newip, port);
                } catch (IOException e1) {
                    e1.printStackTrace();
                }

1. Download google-services.json file, and put the file in the path (Project - App) described on the screen.

2. Set build.gradle(Project), Set build.gradle(App), then Sync

When the Firebase is set, the location of the bridge where the current camera is installed is transmitted to the database whenever a projection is detected.

1. Run smartphone application.

2. Place the camera in a place where it is assumed to be on the bridge.

3. Detect a person crossing from left to right on the line.

4. Remember, people only sense when they cross from left to right of the line.

5. Then, the app sends a picture and bounding box information of falling moment to the connected server.