如何将mask rcnn预测的mask(二值图)转化为标注信息（segmentation）

 说明：该代码是整体代码中的一段，主要功能是将mask_info(bool类型的二维向量)，转化为需要标注的segmentation.

contours, hierarchy = cv2.findContours(mask_info.astype(np.uint8),cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)
xy_list = []
for contour in contours:
    xy = contour[:,0,:].ravel().tolist()
    xy_list.append(xy)

obj_info['area'] = (x2-x1)*(y2-y1)
obj_info['bbox'] = [float(x1), float(y1), float(x2-x1), float(y2-y1)]
obj_info['cate'] = cate
obj_info['segmentation'] = xy_list

  参考文章：博客