parallel_inference.txt

public class Obj_Detect {
    public static void main(String[] args) throws Exception {
        //org.openjdk.jmh.Main.main(args);
        Model model = ModelSerializer.restoreComputationGraph("/home/project/data/tiny_yolov2.zip",false);
        ParallelInference parallelInference = new ParallelInference.Builder(model).inferenceMode(InferenceMode.BATCHED).batchLimit(8).workers(4).build(); //.queueLimit(32)

        new Thread(() -> {
            try {
                new ObjectDetectorFromVideo(parallelInference).startRealTimeVideoDetection("/home/project/videos/adisys_cam_1.mp4");
                );
            } catch (Exception e) {
                e.printStackTrace();
            }
        }).start();
      }
 }

public class ObjectDetectorFromVideo {

    ParallelInference parallelInference;
    public static final double[][] DEFAULT_PRIOR_BOXES = {{0.57273, 0.677385}, {1.87446, 2.06253}, {3.33843, 5.47434}, {7.88282, 3.52778}, {9.77052, 9.16828}};

    public ObjectDetectorFromVideo(ParallelInference parallelInference) throws IOException {
        this.parallelInference = parallelInference;
    }

    public  void startRealTimeVideoDetection(String videoFileName) throws java.lang.Exception { //, TinyYoloModel model
        String windowName = videoFileName;
        FFmpegFrameGrabber frameGrabber = new FFmpegFrameGrabber(videoFileName);
        frameGrabber.start();
        Mat[] v = new Mat[1];
        Frame frame;
        double frameRate = frameGrabber.getFrameRate();
        int num_frames = frameGrabber.getLengthInFrames();
        System.out.println("The inputted video clip has " + frameGrabber.getLengthInFrames() + " frames");
        System.out.println("The inputted video clip has frame rate of " + frameRate);
        long start_time = System.currentTimeMillis();
        NativeImageLoader nil = new NativeImageLoader();
        List<DetectedObject> predictedObjects;

        try {
            for(int i = 1; i < frameGrabber.getLengthInFrames(); i+=1) { //frameGrabber.getLengthInFrames() (int)frameRate
                frameGrabber.setFrameNumber(i);
                frame = frameGrabber.grab();
                v[0] = new OpenCVFrameConverter.ToMat().convert(frame);
                INDArray result = parallelInference.output(prepareImage(v[0],224,224));
                predictedObjects = YoloUtils.getPredictedObjects(Nd4j.create(DEFAULT_PRIOR_BOXES),result,0.5,0);
            }
        } catch (IOException e) {
            e.printStackTrace();
        } finally {
            frameGrabber.stop();
        }
        frameGrabber.close();
    }

    INDArray prepareImage(Mat file, int width, int height) throws IOException {
        NativeImageLoader loader = new NativeImageLoader(height, width, 3);
        ImagePreProcessingScaler imagePreProcessingScaler = new ImagePreProcessingScaler(0, 1);
        INDArray indArray = loader.asMatrix(file);
        imagePreProcessingScaler.transform(indArray);
        return indArray;
    }
}