Run tracking without GUI

[jama1017]

Hi!

Thanks again for this amazing work. I wonder if I can run tracking on a video without invoking the GUI? I know I can pass in an initial bounding box with --optional_box, but even then the GUI still shows up. Is there a way to turn it off?

[moooises]

Hi,
I had a similar problem, because I was executing this on Google Colab.
I leave you here a modified code of run_video_generic without the GUI.
This version save the result in a .avi video. I hope it can help you.

 def run_video_generic(self, debug=None, visdom_info=None, videofilepath=None, optional_box=None, save_results=False):
        """Run the tracker with the webcam or a provided video file.
        args:
            debug: Debug level.
        """
        fourcc = cv.VideoWriter_fourcc(*'DIVX')

        params = self.get_parameters()

        debug_ = debug
        if debug is None:
            debug_ = getattr(params, 'debug', 0)
        params.debug = debug_

        params.tracker_name = self.name
        params.param_name = self.parameter_name

        self._init_visdom(visdom_info, debug_)

        multiobj_mode = getattr(params, 'multiobj_mode', getattr(self.tracker_class, 'multiobj_mode', 'default'))

        if multiobj_mode == 'default':
            tracker = self.create_tracker(params)
            if hasattr(tracker, 'initialize_features'):
                tracker.initialize_features()
        elif multiobj_mode == 'parallel':
            tracker = MultiObjectWrapper(self.tracker_class, params, self.visdom, fast_load=True)
        else:
            raise ValueError('Unknown multi object mode {}'.format(multiobj_mode))

        class UIControl:
            def __init__(self):
                self.mode = 'init'  # init, select, track
                self.target_tl = (-1, -1)
                self.target_br = (-1, -1)
                self.new_init = False

            def mouse_callback(self, event, x, y, flags, param):
                if event == cv.EVENT_LBUTTONDOWN and self.mode == 'init':
                    self.target_tl = (x, y)
                    self.target_br = (x, y)
                    self.mode = 'select'
                elif event == cv.EVENT_MOUSEMOVE and self.mode == 'select':
                    self.target_br = (x, y)
                elif event == cv.EVENT_LBUTTONDOWN and self.mode == 'select':
                    self.target_br = (x, y)
                    self.mode = 'init'
                    self.new_init = True

            def get_tl(self):
                return self.target_tl if self.target_tl[0] < self.target_br[0] else self.target_br

            def get_br(self):
                return self.target_br if self.target_tl[0] < self.target_br[0] else self.target_tl

            def get_bb(self):
                tl = self.get_tl()
                br = self.get_br()

                bb = [min(tl[0], br[0]), min(tl[1], br[1]), abs(br[0] - tl[0]), abs(br[1] - tl[1])]
                return bb

        ui_control = UIControl()

        frame_number = 0

        if videofilepath is not None:
            assert os.path.isfile(videofilepath), "Invalid param {}".format(videofilepath)
            ", videofilepath must be a valid videofile"
            cap = cv.VideoCapture(videofilepath)
            ret, frame = cap.read()
            frame_number += 1
            #cv.imshow(display_name, frame)
        else:
            cap = cv.VideoCapture(0)
            
        resolution=frame.shape
        print(resolution)

        wframe=resolution[0]
        hframe=resolution[1]
        print(wframe)
        print(hframe)
        imgname2=videofilepath.split('/')[-1]

        videoWriter = cv.VideoWriter('output'+imgname2.split('.')[0]+'.avi', fourcc, 20.0, (hframe,wframe))


        next_object_id = 1
        sequence_object_ids = []
        prev_output = OrderedDict()
        output_boxes = OrderedDict()

        if optional_box is not None:
            assert isinstance(optional_box, (list, tuple))
            assert len(optional_box) == 4, "valid box's format is [x,y,w,h]"

            out = tracker.initialize(frame, {'init_bbox': OrderedDict({next_object_id: optional_box}),
                                       'init_object_ids': [next_object_id, ],
                                       'object_ids': [next_object_id, ],
                                       'sequence_object_ids': [next_object_id, ]})

            prev_output = OrderedDict(out)

            output_boxes[next_object_id] = [optional_box, ]
            sequence_object_ids.append(next_object_id)
            next_object_id += 1

        # Wait for initial bounding box if video!
        paused = videofilepath is not None
        paused=False
        while True:

            if not paused:
                # Capture frame-by-frame
                ret, frame = cap.read()
                frame_number += 1
                if frame is None:
                    break

            frame_disp = frame.copy()

            info = OrderedDict()
            info['previous_output'] = prev_output

            if ui_control.new_init:
                ui_control.new_init = False
                init_state = ui_control.get_bb()

                info['init_object_ids'] = [next_object_id, ]
                info['init_bbox'] = OrderedDict({next_object_id: init_state})
                sequence_object_ids.append(next_object_id)
                if save_results:
                    output_boxes[next_object_id] = [init_state, ]
                next_object_id += 1

            # Draw box
            if ui_control.mode == 'select':
                cv.rectangle(frame_disp, ui_control.get_tl(), ui_control.get_br(), (255, 0, 0), 2)

            if len(sequence_object_ids) > 0:
                info['sequence_object_ids'] = sequence_object_ids
                out = tracker.track(frame, info)
                prev_output = OrderedDict(out)

                if 'segmentation' in out:
                    frame_disp = overlay_mask(frame_disp, out['segmentation'])
                    mask_image = np.zeros(frame_disp.shape, dtype=frame_disp.dtype)

                    if save_results:
                        mask_image = overlay_mask(mask_image, out['segmentation'])
                        if not os.path.exists(self.results_dir):
                            os.makedirs(self.results_dir)
                        cv.imwrite(self.results_dir + f"seg_{frame_number}.jpg", mask_image)

                if 'target_bbox' in out:
                    for obj_id, state in out['target_bbox'].items():
                        state = [int(s) for s in state]
                        cv.rectangle(frame_disp, (state[0], state[1]), (state[2] + state[0], state[3] + state[1]),
                                     _tracker_disp_colors[obj_id], 5)
                        if save_results:
                            output_boxes[obj_id].append(state)

            videoWriter.write(frame_disp)
            print(f"Frame: {frame_number}")

            # Put text
            font_color = (255, 255, 255)
            msg = "Select target(s). Press 'r' to reset or 'q' to quit."
            #cv.rectangle(frame_disp, (5, 5), (630, 40), (50, 50, 50), -1)
            #cv.putText(frame_disp, msg, (10, 30), cv.FONT_HERSHEY_COMPLEX_SMALL, 1, font_color, 2)

            if videofilepath is not None:
                msg = "Press SPACE to pause/resume the video."
                #cv.rectangle(frame_disp, (5, 50), (530, 90), (50, 50, 50), -1)
                #cv.putText(frame_disp, msg, (10, 75), cv.FONT_HERSHEY_COMPLEX_SMALL, 1, font_color, 2)

        # When everything done, release the capture
        cap.release()
        cv.destroyAllWindows()
        videoWriter.release()

        if save_results:
            if not os.path.exists(self.results_dir):
                os.makedirs(self.results_dir)
            video_name = "webcam" if videofilepath is None else Path(videofilepath).stem
            base_results_path = os.path.join(self.results_dir, 'video_{}'.format(video_name))
            print(f"Save results to: {base_results_path}")
            for obj_id, bbox in output_boxes.items():
                tracked_bb = np.array(bbox).astype(int)
                bbox_file = '{}_{}.txt'.format(base_results_path, obj_id)
                np.savetxt(bbox_file, tracked_bb, delimiter='\t', fmt='%d')

[jama1017]

Thanks so much @moooises ! This is very helpful

[hphnngcquan]

Hi @moooises, can i track multiple object using this script? and How can i do it? Thank you

[moooises]

Hi @hphnngcquan
You need to modify a bit the script.
In the block where it access the optional box, change it for something like this.

optional_boxes=[(670, 366, 50, 50),(513, 255, 71, 65)]

if optional_box is not None:
    assert isinstance(optional_box, (list, tuple))
    assert len(optional_box) == 4, "valid box's format is [x,y,w,h]"
    boxes=OrderedDict()
    ids=[]
    for i,box in enumerate(optional_boxes):
        i+=1
        boxes.update({i:box})
        ids.append(i)
        output_boxes[i]=[box,]
        sequence_object_ids.append(i)

    out = tracker.initialize(frame, {'init_bbox': boxes,
                            'init_object_ids': ids,
                            'object_ids': ids,
                            'sequence_object_ids': ids})
    prev_output = OrderedDict(out)

In the optional_boxes list, place the coordinates of the objects you want to track.

[hphnngcquan]

Many thanks @moooises

[hphnngcquan]

@moooises
If you have experience about how to track multiple objects with run_sequence() function, pls enlighten me
Appreciate your help!!

[moooises]

Sorry, I have just worked with the run_video_generic function.

[hphnngcquan]

many thanks