tensorflow 加载多个pb模型节点冲突解决

           使用多个pb模型时，有时会遇到模型间节点相同的情况，造成图加载冲突，无法输出正确结果，更改节点信息较为麻烦，因此创建新图来解决

 下面是直接使用默认图tf.get_default_graph()， 如果模型间无冲突可使用。

    def __init__(self, graph_path, target_size=(300, 300), confidence_thresh=0.9):
        self.target_size = target_size
        self.confidence_threshold = confidence_thresh
        # load graph
        with tf.gfile.FastGFile(graph_path, 'rb') as f:
            graph_def = tf.GraphDef()
            graph_def.ParseFromString(f.read())
            _ = tf.import_graph_def(graph_def, name='')

        self.graph = tf.get_default_graph()
        gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.03)
        self.fire_Detecction_sess = tf.Session(graph=self.graph, config=tf.ConfigProto(gpu_options=gpu_options))

        self.input = self.fire_Detecction_sess.graph.get_tensor_by_name("input_1:0")
        self.output = self.fire_Detecction_sess.graph.get_tensor_by_name("output_1:0")

 下面是模型间有节点同名冲突时，可用tf.Graph()为当前模型创建一个新图来解决问题。

    def __init__(self, graph_path, target_size=(300, 300), confidence_thresh=0.9):
        self.target_size = target_size
        self.confidence_threshold = confidence_thresh

        self.graph = tf.Graph()
        self.graph_def = tf.GraphDef()
        with tf.gfile.FastGFile(graph_path, 'rb') as f:
            self.graph_def.ParseFromString(f.read())
        with self.graph.as_default():
            tf.import_graph_def(self.graph_def, name='')

        gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.03)
        self.fire_Detecction_sess = tf.Session(graph=self.graph, config=tf.ConfigProto(gpu_options=gpu_options))

        self.input = self.fire_Detecction_sess.graph.get_tensor_by_name("input_1:0")
        self.output = self.fire_Detecction_sess.graph.get_tensor_by_name("output_1:0")