Description
I have fine-tuned Tensorflow2 object detection API models using my custom dataset. To be more precise, I am using “ssd_resnet50_v1_fpn_640x640_coco17_tpu-8” and all the related codes in this github repo “GitHub - tensorflow/models: Models and examples built with TensorFlow”. After training the models, I used this module “models/exporter_main_v2.py at master · tensorflow/models · GitHub” to convert the output checkpoints to SavedModel format. Finally, I used this github repo “GitHub - tensorflow/tensorrt: TensorFlow/TensorRT integration” to convert my SavedModel to trt. The model size before conversion is about 13MB and after it it is 402MB. Would you please help me to figure out what the problem is?
Environment
TensorRT Version: TensorRT 7.2.2
GPU Type: GeForce RTX 3090
Nvidia Driver Version: 455.23.04
CUDA Version: cuda11.1.0.024
CUDNN Version: * NVIDIA cuDNN 8.1.0]
Operating System + Version: Ubuntu 20.04
Python Version (if applicable): Python 3.8
TensorFlow Version (if applicable): 2.4.0
PyTorch Version (if applicable):
Baremetal or Container (if container which image + tag): nvcr.io/nvidia/tensorflow:21.03-tf2-py3
Relevant Files
Steps To Reproduce
Use the attached Jupiter notebook step by step