def extract(cls, node): encoding_map = {0: 'corner', 1: 'center'} center_point_box = onnx_attr(node, 'center_point_box', 'i', default=0) NonMaxSuppression.update_node_stat(node, {'sort_result_descending': 0, 'output_type': np.int64, 'box_encoding': encoding_map[center_point_box]}) return cls.enabled
def extract(cls, node): attrs = { 'sort_result_descending': 1, 'box_encoding': 'corner', 'output_type': np.int32 } NonMaxSuppression.update_node_stat(node, attrs) return cls.enabled
def extract(cls, node): pad_to_max_output_size = node.pb.attr["pad_to_max_output_size:"].b if not pad_to_max_output_size: log.warning('The attribute "pad_to_max_output_size" of node {} is equal to False which is not supported.' 'Forcing it to be equal to True'.format(node.soft_get('name'))) attrs = {'sort_result_descending': 1, 'box_encoding': 'corner', 'output_type': np.int32} NonMaxSuppression.update_node_stat(node, attrs) return cls.enabled
def extract(cls, node): attrs = {'sort_result_descending': 1, 'center_point_box': 0, 'output_type': np.int32} NonMaxSuppression.update_node_stat(node, attrs) return cls.enabled