This repository provides tools and resources to benchmark TensorFlow Lite models on various hardware platforms specialy arm based embedded systems, making it easier for developers and researchers to measure the performance of their models.
TensorFlow Lite is an optimized machine learning framework for mobile and embedded devices. Benchmarking TensorFlow Lite models is crucial for understanding their runtime performance on various hardware, including CPUs, GPUs, and accelerators like Edge TPUs. This repository is designed to help you evaluate and compare the performance of your TensorFlow Lite models across different devices and configurations.
Follow these steps to set up the benchmarking environment.
- Python 3.6 or higher
- TensorFlow Lite (TFLite) library
- TensorFlow (optional, for TensorFlow models)
- Supported hardware platforms (e.g., Android devices, Edge TPUs, Raspberry Pi)
-
Clone this repository to your local machine:
git clone https://github.com/Mohammadakhavan75/tflite-benchmark.git
1.1 Installing required packages
1.1.1 If you're running Debian Linux or a derivative of Debian please use this:
echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
sudo apt-get update
sudo apt-get install python3-tflite-runtime
1.1.1 For other systems you can use:
pip install --extra-index-url https://google-coral.github.io/py-repo/ tflite_runtime
pip install -r requirements.txt