Unable to visualise point cloud in RViz even with header.frame_id selected

edit

Hi all, I very new to ROS, so please bear with me. I am trying to render sample point cloud in the RViz, but for some reason it just won't show. I have set fixed frame to map which is the value of header.frame_id. rostopic echo on the point cloud topic returns data, so my take is that the point cloud is correctly generated but cannot be properly displayed in rviz. Here's the code for publishing the point cloud:

void Node::PublishMapPoints (std::vector<ORB_SLAM2::MapPoint*> map_points) {
  sensor_msgs::PointCloud2 cloud = MapPointsToPointCloud (map_points);
  map_points_publisher_.publish (cloud);
}

sensor_msgs::PointCloud2 Node::MapPointsToPointCloud (std::vector<ORB_SLAM2::MapPoint*> map_points) {
  if (map_points.size() == 0) {
    std::cout << "Map point vector is empty!" << std::endl;
  }

  sensor_msgs::PointCloud2 cloud;

  const int num_channels = 3; // x y z

  cloud.header.stamp = current_frame_time_;
  cloud.header.frame_id = "map";
  cloud.height = 1;
  cloud.width = map_points.size();
  cloud.is_bigendian = false;
  cloud.is_dense = true;
  cloud.point_step = num_channels * sizeof(float);
  cloud.row_step = cloud.point_step * cloud.width;
  cloud.fields.resize(num_channels);

  std::string channel_id[] = { "x", "y", "z"};
  for (int i = 0; i<num_channels; i++) {
    cloud.fields[i].name = channel_id[i];
    cloud.fields[i].offset = i * sizeof(float);
    cloud.fields[i].count = 1;
    cloud.fields[i].datatype = sensor_msgs::PointField::FLOAT32;
  }

  cloud.data.resize(cloud.row_step * cloud.height);

    unsigned char *cloud_data_ptr = &(cloud.data[0]);

  float data_array[num_channels];
  for (unsigned int i=0; i<cloud.width; i++) {
    if (map_points.at(i)->nObs >= min_observations_per_point_) {
      data_array[0] = map_points.at(i)->GetWorldPos().at<float> (2); //x. Do the transformation by just reading at the position of z instead of x
      data_array[1] = -1.0* map_points.at(i)->GetWorldPos().at<float> (0); //y. Do the transformation by just reading at the position of x instead of y
      data_array[2] = -1.0* map_points.at(i)->GetWorldPos().at<float> (1); //z. Do the transformation by just reading at the position of y instead of z
      //TODO dont hack the transformation but have a central conversion function for MapPointsToPointCloud and TransformFromMat

      memcpy(cloud_data_ptr+(i*cloud.point_step), data_array, num_channels*sizeof(float));
    }
  }

  return cloud;
}

Here's my RViz:

Any help would be much appreciated.