TensorFlow-Slim image classification model library

Contacts

Citation

Table of contents

Installation

Installing latest version of TF-slim

Installing the TF-slim image models library

Preparing the datasets

Downloading and converting to TFRecord format

Creating a TF-Slim Dataset Descriptor.

An automated script for processing ImageNet data.

Pre-trained Models

Training a model from scratch.

TensorBoard

Fine-tuning a model from an existing checkpoint

Evaluating performance of a model

Exporting the Inference Graph

Freezing the exported Graph

Run label image in C++

Troubleshooting

The model runs out of CPU memory.

The model runs out of GPU memory.

The model training results in NaN's.

The ResNet and VGG Models have 1000 classes but the ImageNet dataset has 1001

I wish to train a model with a different image size.

What hardware specification are these hyper-parameters targeted for?