Tensorflow-benchmarks

Tensorflow model inference Benchmarks via slim model zoo

Introduction

This repository contains machine learning models implemented in TensorFlow. It was developed for easily test network architecture and performance via slim model define. The models are maintained by their respective authors.

Requirement

• Tensorflow
• opencv

Installation

1. Clone this repository

   $> git clone https://github.com/LemonAniLabs/Tensorflow-benchmarks.git

2. Download the pre-train weight from tf-slim

   # Download resnet_v1_50 pre-train weight
   $> wget http://download.tensorflow.org/models/resnet_v1_50_2016_08_28.tar.gz

Run the benchmark

    $> python resnet_v1_50_test.py
    # That's it

Optional arguments:

  -h, --help            show this help message and exit
  -t TIMES, --TIMES TIMES
                        Times of iteration
  -o, --OFFICIAL        Use Slim.Net from tensorflow contrib

Features

• Training a model from scratch
• Fine-tuning a model from an existing checkpoint