def _calculate_num_points_in_gt(data_path,
infos,
relative_path,
remove_outside=True,
num_features=4):
for info in infos:
if relative_path:
v_path = str(pathlib.Path(data_path) / info["velodyne_path"])
else:
v_path = info["velodyne_path"]
points_v = np.fromfile(v_path, dtype=np.float32,
count=-1).reshape([-1, num_features])
rect = info['calib/R0_rect']
Trv2c = info['calib/Tr_velo_to_cam']
P2 = info['calib/P2']
if remove_outside:
points_v = remove_outside_points(points_v, rect, Trv2c, P2,
info["img_shape"])
# points_v = points_v[points_v[:, 0] > 0]
annos = info['annos']
num_obj = len([n for n in annos['name'] if n != 'DontCare'])
# annos = kitti.filter_kitti_anno(annos, ['DontCare'])
dims = annos['dimensions'][:num_obj]
loc = annos['location'][:num_obj]
rots = annos['rotation_y'][:num_obj]
gt_boxes_camera = np.concatenate([loc, dims, rots[..., np.newaxis]],
axis=1)
gt_boxes_lidar = box_camera_to_lidar(gt_boxes_camera, rect, Trv2c)
indices = points_in_rbbox(points_v[:, :3], gt_boxes_lidar)
num_points_in_gt = indices.sum(0)
num_ignored = len(annos['dimensions']) - num_obj
num_points_in_gt = np.concatenate(
[num_points_in_gt, -np.ones([num_ignored])])
annos["num_points_in_gt"] = num_points_in_gt.astype(np.int32)
def prepare_train_img(self, idx):
sample_id = self.sample_ids[idx]
# load image
img = mmcv.imread(osp.join(self.img_prefix, '%06d.png' % sample_id))
img, img_shape, pad_shape, scale_factor = self.img_transform(
img, 1, False)
objects = read_label(
osp.join(self.label_prefix, '%06d.txt' % sample_id))
calib = Calibration(osp.join(self.calib_prefix,
'%06d.txt' % sample_id))
gt_bboxes = [
object.box3d for object in objects
if object.type not in ["DontCare"]
]
gt_bboxes = np.array(gt_bboxes, dtype=np.float32)
gt_types = [
object.type for object in objects
if object.type not in ["DontCare"]
]
# transfer from cam to lidar coordinates
if len(gt_bboxes) != 0:
gt_bboxes[:, :3] = project_rect_to_velo(gt_bboxes[:, :3], calib)
img_meta = dict(img_shape=img_shape, sample_idx=sample_id, calib=calib)
data = dict(img=to_tensor(img), img_meta=DC(img_meta, cpu_only=True))
if self.anchors is not None:
data['anchors'] = DC(to_tensor(self.anchors.astype(np.float32)))
if self.with_mask:
NotImplemented
if self.with_point:
points = read_lidar(
osp.join(self.lidar_prefix, '%06d.bin' % sample_id))
if self.augmentor is not None and self.test_mode is False:
sampled_gt_boxes, sampled_gt_types, sampled_points = self.augmentor.sample_all(
gt_bboxes, gt_types)
assert sampled_points.dtype == np.float32
gt_bboxes = np.concatenate([gt_bboxes, sampled_gt_boxes])
gt_types = gt_types + sampled_gt_types
assert len(gt_types) == len(gt_bboxes)
# to avoid overlapping point (option)
masks = points_in_rbbox(points, sampled_gt_boxes)
#masks = points_op_cpu.points_in_bbox3d_np(points[:,:3], sampled_gt_boxes)
points = points[np.logical_not(masks.any(-1))]
# paste sampled points to the scene
points = np.concatenate([sampled_points, points], axis=0)
# select the interest classes
selected = [
i for i in range(len(gt_types))
if gt_types[i] in self.class_names
]
gt_bboxes = gt_bboxes[selected, :]
gt_types = [
gt_types[i] for i in range(len(gt_types))
if gt_types[i] in self.class_names
]
# force van to have same label as car
gt_types = ['Car' if n == 'Van' else n for n in gt_types]
gt_labels = np.array(
[self.class_names.index(n) + 1 for n in gt_types],
dtype=np.int64)
self.augmentor.noise_per_object_(gt_bboxes, points, num_try=100)
gt_bboxes, points = self.augmentor.random_flip(gt_bboxes, points)
gt_bboxes, points = self.augmentor.global_rotation(
gt_bboxes, points)
gt_bboxes, points = self.augmentor.global_scaling(
gt_bboxes, points)
if isinstance(self.generator, VoxelGenerator):
#voxels, coordinates, num_points = self.generator.generate(points)
voxel_size = self.generator.voxel_size
pc_range = self.generator.point_cloud_range
grid_size = self.generator.grid_size
keep = points_op_cpu.points_bound_kernel(points, pc_range[:3],
pc_range[3:])
voxels = points[keep, :]
coordinates = (
(voxels[:, [2, 1, 0]] -
np.array(pc_range[[2, 1, 0]], dtype=np.float32)) /
np.array(voxel_size[::-1], dtype=np.float32)).astype(np.int32)
num_points = np.ones(len(keep)).astype(np.int32)
data['voxels'] = DC(to_tensor(voxels.astype(np.float32)))
data['coordinates'] = DC(to_tensor(coordinates))
data['num_points'] = DC(to_tensor(num_points))
if self.anchor_area_threshold >= 0 and self.anchors is not None:
dense_voxel_map = sparse_sum_for_anchors_mask(
coordinates, tuple(grid_size[::-1][1:]))
dense_voxel_map = dense_voxel_map.cumsum(0)
dense_voxel_map = dense_voxel_map.cumsum(1)
anchors_area = fused_get_anchors_area(dense_voxel_map,
self.anchors_bv,
voxel_size, pc_range,
grid_size)
anchors_mask = anchors_area > self.anchor_area_threshold
data['anchors_mask'] = DC(
to_tensor(anchors_mask.astype(np.uint8)))
# filter gt_bbox out of range
bv_range = self.generator.point_cloud_range[[0, 1, 3, 4]]
mask = filter_gt_box_outside_range(gt_bboxes, bv_range)
gt_bboxes = gt_bboxes[mask]
gt_labels = gt_labels[mask]
else:
NotImplementedError
# skip the image if there is no valid gt bbox
if len(gt_bboxes) == 0:
return None
# limit rad to [-pi, pi]
gt_bboxes[:, 6] = limit_period(gt_bboxes[:, 6],
offset=0.5,
period=2 * np.pi)
if self.with_label:
data['gt_labels'] = DC(to_tensor(gt_labels))
data['gt_bboxes'] = DC(to_tensor(gt_bboxes))
return data
def create_groundtruth_database(data_path,
info_path=None,
used_classes=None,
database_save_path=None,
db_info_save_path=None,
relative_path=True,
lidar_only=False,
bev_only=False,
coors_range=None):
root_path = pathlib.Path(data_path)
if info_path is None:
info_path = root_path / 'kitti_infos_train.pkl'
if database_save_path is None:
database_save_path = root_path / 'gt_database'
else:
database_save_path = pathlib.Path(database_save_path)
if db_info_save_path is None:
db_info_save_path = root_path / "kitti_dbinfos_train.pkl"
database_save_path.mkdir(parents=True, exist_ok=True)
with open(info_path, 'rb') as f:
kitti_infos = pickle.load(f)
all_db_infos = {}
if used_classes is None:
used_classes = list(kitti.get_classes())
used_classes.pop(used_classes.index('DontCare'))
for name in used_classes:
all_db_infos[name] = []
group_counter = 0
for info in prog_bar(kitti_infos):
velodyne_path = info['velodyne_path']
if relative_path:
# velodyne_path = str(root_path / velodyne_path) + "_reduced"
velodyne_path = str(root_path / velodyne_path)
num_features = 4
if 'pointcloud_num_features' in info:
num_features = info['pointcloud_num_features']
points = np.fromfile(velodyne_path, dtype=np.float32,
count=-1).reshape([-1, num_features])
image_idx = info["image_idx"]
rect = info['calib/R0_rect']
P2 = info['calib/P2']
Trv2c = info['calib/Tr_velo_to_cam']
if not lidar_only:
points = remove_outside_points(points, rect, Trv2c, P2,
info["img_shape"])
annos = info["annos"]
names = annos["name"]
bboxes = annos["bbox"]
difficulty = annos["difficulty"]
gt_idxes = annos["index"]
num_obj = np.sum(annos["index"] >= 0)
rbbox_cam = kitti.anno_to_rbboxes(annos)[:num_obj]
rbbox_lidar = box_camera_to_lidar(rbbox_cam, rect, Trv2c)
if bev_only: # set z and h to limits
assert coors_range is not None
rbbox_lidar[:, 2] = coors_range[2]
rbbox_lidar[:, 5] = coors_range[5] - coors_range[2]
group_dict = {}
group_ids = np.full([bboxes.shape[0]], -1, dtype=np.int64)
if "group_ids" in annos:
group_ids = annos["group_ids"]
else:
group_ids = np.arange(bboxes.shape[0], dtype=np.int64)
point_indices = points_in_rbbox(points, rbbox_lidar)
for i in range(num_obj):
filename = f"{image_idx}_{names[i]}_{gt_idxes[i]}.bin"
filepath = database_save_path / filename
gt_points = points[point_indices[:, i]]
gt_points[:, :3] -= rbbox_lidar[i, :3]
with open(filepath, 'w') as f:
gt_points.tofile(f)
if names[i] in used_classes:
if relative_path:
db_path = str(database_save_path.stem + "/" + filename)
else:
db_path = str(filepath)
db_info = {
"name": names[i],
"path": db_path,
"image_idx": image_idx,
"gt_idx": gt_idxes[i],
"box3d_lidar": rbbox_lidar[i],
"num_points_in_gt": gt_points.shape[0],
"difficulty": difficulty[i],
# "group_id": -1,
# "bbox": bboxes[i],
}
local_group_id = group_ids[i]
# if local_group_id >= 0:
if local_group_id not in group_dict:
group_dict[local_group_id] = group_counter
group_counter += 1
db_info["group_id"] = group_dict[local_group_id]
if "score" in annos:
db_info["score"] = annos["score"][i]
all_db_infos[names[i]].append(db_info)
for k, v in all_db_infos.items():
print(f"load {len(v)} {k} database infos")
with open(db_info_save_path, 'wb') as f:
pickle.dump(all_db_infos, f)