search

mo12896 / facial-analysis-system

This repository contains the code and documentation for my master's thesis, which focuses on predicting well-being from visual data using novel AI tools for face detection, tracking, re-identification, facial emotion recognition and head pose estimation.
MIT License
2 stars 0 forks source link

Solve TensorRT Bug #17

Closed mo12896 closed 1 year ago

mo12896 commented 1 year ago
mo12896 commented 1 year ago

This is a warning message from TensorFlow indicating that the TensorRT libraries could not be loaded. TensorRT is a high-performance deep learning inference library from Nvidia, and is used to optimize TensorFlow models for deployment on Nvidia GPUs.

The message states that two TensorRT libraries (libnvinfer.so.7 and libnvinfer_plugin.so.7) could not be opened because the file does not exist. This suggests that TensorRT may not be installed properly on your system, or that the libraries are not located in a directory that is in the LD_LIBRARY_PATH environment variable.

To resolve this issue, you can either install TensorRT and make sure the libraries are in a directory that is in the LD_LIBRARY_PATH, or you can reconfigure TensorFlow to not use TensorRT.

mo12896 commented 1 year ago

Install the whole thing, but not from pip (https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html)

mo12896 commented 1 year ago

-> Ignore, when TensorRT is not needed, otherwise downgrade to TensorRT 7.X using python=3.8 in emorec env (https://github.com/tensorflow/tensorflow/issues/57679)