Converts a single box_3d to box_4c Args: box_3d: box_3d (6,) ground_plane: ground plane coefficients (4,) Returns: b

Computes the 3D bounding box corner positions from box_3d format. This function does not preserve corners order but rather the corners are ro

Computes the 3D bounding box corner positions from Box3D format. The order of corners are preserved during this conversion. Args: bo

Converts a single box_4c to box_3d. The longest midpoint-midpoint length is used to calculate orientation. Points are projected onto the orien

Returns number of boxes held in collection. Returns: a tensor representing the number of boxes held in the collection.

Returns number of boxes held in collection. This number is inferred at graph construction time rather than run-time. Returns:

Turns an ObjectLabel into an box_3d Args: obj_label: ObjectLabel Returns: anchor: 3D box in box_3d format [x, y, z, l, w, h,

Decodes the anchor regression predictions with the anchor. Args: anchors: A numpy array or a tensor of shape [N, 6] repre

Preprocesses the given input. Args: tensor_in: A `Tensor` of shape=(batch_size, height, width, channels) represen

Computes the 3D bounding box projected onto image space. Keyword arguments: obj -- object file to draw bounding box p -- transform ma

Projects an array of 3D boxes into bird's eye view Args: boxes_3d: list of 3d boxes in the format: N x [x, y, z, l, w, h

Projects an array of 3D anchors into bird's eye view Args: anchors: list of anchors in anchor format (N x 6): N x [x, y,

Projects a 3D point cloud to 2D points for plotting :param point_cloud: 3D point cloud (3, N) :param p: Camera matrix (3, 4) :return: p

Projects a box_3d into image space Args: box_3d: single box_3d to project calib_p2: stereo calibration p2 matrix truncat

Projects 3D anchors into image space Args: anchors: list of anchors in anchor format N x [x, y, z, dim_x, dim_y, dim_z]

Hack to convert proto config into an object so repeated fields can be overwritten Args: config: proto config Returns: co

Prunes the boxes in boxlist1 that overlap less than thresh with boxlist2. For each box in boxlist1, we want its IOA to be more than minoverlap wi

Prunes small boxes in the boxlist which have a side smaller than min_side. Args: boxlist: BoxList holding N boxes. min_side: Minimum

Reads in Calibration file from Kitti Dataset. Keyword Arguments: ------------------ calib_dir : Str Directory of the cali

Reads in Disparity file from Kitti Dataset. Keyword Arguments: ------------------ calib_dir : Str Directo

Reads in label data file from Kitti Dataset. Returns: obj_list -- List of instances of class ObjectLabel. Keyword arguments: label_d

Reads in PointCloud from Kitti Dataset. Keyword Arguments: ------------------ velo_dir : Str Directory of

Helper function to reorder image corners. This reorders the corners from [x1, y1, x2, y2] to [y1, x1, y2, x2] which is required by the tf.cro

Runs the kitti native code script.

Runs the kitti native code script.

Converts a set of network predictions into text files required for KITTI evaluation.

Converts a set of network predictions into text files required for KITTI evaluation.

Sets the value of a field. Updates the field of a box_list with a given value. Args: field: (string) name of the field to

Helper function to set up log directories and summary handlers. Args: model_config: Model protobuf configuration sess : A

Computes the pairwise squared distances between box corners. This op treats each box as if it were a point in a 4d Euclidean space and comput

Helper function to return the checkpoint index number. Args: checkpoint_dir: Path directory of the checkpoints Returns: chec

image orientation

image orientation

image orientation

false negatives

false negatives

false negatives

Encodes the anchor regression predictions with the ground truth. This function assumes the ground_truth tensor has been arranged in a way

Converts angle unit vectors into orientation angle representation. e.g. [0.717, 0.717] -> 45, [0, 1] -> 90 Args: angle_vectors_t

Returns the diagonal length of box_3d Args: boxes_3d: An tensor of shape (N x 7) of boxes in box_3d format. Returns: Diagona

Converts a box_3d tensor to anchor format by ortho rotating it. This is similar to 'box_3d_to_anchor' above however it takes a tensor as input

Vectorized conversion of box_3d to box_4c tensors Args: boxes_3d: Tensor of boxes_3d (N, 7) ground_plane: Tensor ground plane coe

Computes the 3D bounding box corner positions from box_3d format. This function does not preserve corners order during conversion from box_3d

Computes the 3D bounding box corner positions from Box3D format. The order of corners are preserved during this conversion. Args: bo

Vectorized box_4c to box_3d conversion Args: boxes_4c: Tensor of boxes_4c (N, 10) ground_plane: Tensor of ground plane coefficien

Calculates box_4c offsets to regress to ground truth Args: boxes_4c: boxes_4c to calculate offset for (N, 10) box_4c_gt: box_4c g

Converts corner boxes to corner offsets. It subtracts the ground-truth box corners from the predicted corners and normalizes the offsets by th

Applies box_4c offsets to boxes_4c Args: boxes_4c: boxes_4c to apply offsets to offsets: box_4c offsets to apply Returns:

Converts corner ofsets to box corners. It multiplies the diagonals with the offsets and then adds it back to the box corners. Args:

Converts orientation angles into angle unit vector representation. e.g. 45 -> [0.717, 0.717], 90 -> [0, 1] Args: orientations_te

Projects 3D tensor anchors into image space Args: anchors: a tensor of anchors in the shape [N, 6]. The anchors are in t

Computes approximate 3D IOU between a 3D bounding box 'box' and a list of 3D bounding boxes 'boxes'. All boxes are assumed to be aligned with

Training function for detection models. Args: model: The detection model object. train_config: a train_*pb2 protobuf.

Transpose the coordinate representation in a boxlist. Args: scope: name scope of the function.

Compute 2D IOU between a 2D bounding box 'box' and a list :param box: a numpy array in the form of [x1, y1, x2, y2] where (x1,y1) are image c