key C， V bug

[Larry-Zheng]

First of all, I appreciate your work on VOD annotation. I've been searching for relative github topics. And labelfficient seems to be the only practical solution so far (although the tk gui is really poor). My problem is that, after I switch a label or add a new label, key C and key V break down. Finally, I find that the callback event of key C and key V return negative （x,y） coordinates. I don't know why ，so as a compromise, I add another 2 keys of STATE dict to keep track of mouse move (x,y). I hope you could fix it. Best wishes！

[Larry-Zheng]

def smart_translate_box(self, event=None):
    if event is None : return
    event.x = self.STATE['x_move']
    event.y = self.STATE['y_move']
    mouse_pos = self.get_mouse_pos(event)
    closest_box, distance, inside = self.get_closest_box(mouse_pos)
    # print(mouse_pos,closest_box, distance, inside)
    if closest_box is None:
        return
    prev_det = [Detection(self.class_list[closest_box], self.bbox_list[closest_box])]
    prev_img = np.array(self.img)
    self.next_image(event, load_labels=False)
    cur_img = np.array(self.img)
    tracked = self.track(prev_img, cur_img, prev_det)
    self.load_labels(objects=tracked, load_anyway=True)

[Larry-Zheng]

My system is Windows