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