Return a DetectionList object from the gobs All object are still in the camera frame.

Compute the containing relationship between all pair of bounding boxes. For each mask, subtract the mask of bounding boxes that are contained

Merge the new object to the old object This operation is done in-place

Args: image (np.ndarray): an image of shape (H, W, C) (in BGR order). This is the format used by OpenCV.

Args: init_grid: B X nF X W X H X D X .. feat: B X nF X nPt coords: B X nDims X nPt in [-1, 1] Returns: grid:

Transforms the point cloud into geocentric frame to account for camera elevation and angle Input: XYZ : ...x3

Transforms the point cloud into geocentric frame to account for camera position Input: XYZ : ...x3 cu

Args: image_list (list of np.ndarray): a list of images of shape (H, W, C) (in BGR o

Args: cfg (CfgNode): instance_mode (ColorMode):

Bins points into xy-z bins XYZ_cms is ... x H x W x3 Outputs is ... x map_size x map_size x (len(z_bins)+1)

obstacle map (H, W): 1 means occupied, 0 means free start (2, ): planning start top_k_goal (k, 2): fmm generated distance field top k sho

Color the point cloud of each detection by the most common class

The input feature should be of shape (D, ), a one-row vector This is mostly for backward compatibility