open-closed-eye-0001

## Use Case and High-Level Description

Fully convolutional network for recognition of eye state ('open', 'closed').

## Specification

Metric Value
Source framework PyTorch*
GFlops 0.0014
MParams 0.0113
Accuracy 95.84%

## Dataset

MRL Eye Dataset is used for training and validation (each 10th image is used for test).

## Inputs

Image, name: input.1, shape: 1, 3, 32, 32 in 1, C, H, W format, where:

• C - channel
• H - height
• W - width

Expected color order is BGR.

## Outputs

Name: 19, shape: 1, 2, 1, 1 - Softmax output across 2 type classes: [open, closed]

You can download models and if necessary convert them into Inference Engine format using the Model Downloader and other automation tools as shown in the examples below.