Skipping full serialization of Keras layer because it is not built.

[SJavad]

i trained SSD-MobileNet on custom dataset and now I want to convert the model to TensorRT model for improve performance and for Inference on Nvidia Jetson TX1 Board.

I Have: -Ubuntu 20.14 LTS -Tensorflow 2.11.0 -Python 3.8.10

when I run the:

import tensorflow as tf
tf.saved_model.save(model_dir , out_dir)

I wil get this error:

WARNING:tensorflow:Skipping full serialization of Keras layer <object_detection.meta_architectures.ssd_meta_arch.SSDMetaArch object at 0x7f45581d7340>, because it is not built.
WARNING:tensorflow:Skipping full serialization of Keras layer <object_detection.predictors.convolutional_keras_box_predictor.WeightSharedConvolutionalBoxPredictor object at 0x7f4558265760>, because it is not built.
WARNING:tensorflow:Skipping full serialization of Keras layer <object_detection.models.ssd_mobilenet_v2_fpn_keras_feature_extractor.SSDMobileNetV2FpnKerasFeatureExtractor object at 0x7f4561e31130>, because it is not built.
WARNING:tensorflow:Skipping full serialization of Keras layer <object_detection.predictors.heads.keras_box_head.WeightSharedConvolutionalBoxHead object at 0x7f45582562b0>, because it is not built.
WARNING:tensorflow:Skipping full serialization of Keras layer <object_detection.predictors.heads.keras_class_head.WeightSharedConvolutionalClassHead object at 0x7f4558256ca0>, because it is not built.
WARNING:tensorflow:Skipping full serialization of Keras layer <keras.layers.convolutional.separable_conv2d.SeparableConv2D object at 0x7f4558256460>, because it is not built.
WARNING:tensorflow:Skipping full serialization of Keras layer <keras.layers.convolutional.separable_conv2d.SeparableConv2D object at 0x7f45582650d0>, because it is not built.
WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 2 of 2). These functions will not be directly callable after loading.

how can I fix that ?