Skip to content

Latest commit

 

History

History
 
 

open_closed_eye

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
utils
utils
 
 
README.md
README.md
 
 
demo.py
demo.py
 
 
img.jpg
img.jpg
 
 
init_venv.sh
init_venv.sh
 
 
requirements.txt
requirements.txt
 
 
train.py
train.py
 
 

README.md

Eye State classification

This repository contains training scripts for the eye state classifier.

Setup

Prerequisites

  • Ubuntu* 16.04 / 18.04
  • Python* 3.6
  • PyTorch* 1.4.0
  • OpenVINO™ with Python API

Installation

  1. Create virtual environment:
bash init_venv.sh
  1. Activate virtual environment:
. venv/bin/activate

Training and Evaluation

To train and evaluate net:

python3 train.py \
    --data_root ../../data/open_closed_eye/ \
    --epoch 1 \
    --batch 1 \
    --learning_rate 0.1

Convert to OpenVino

mo.py --input_model model/open_closed_eye_epoch_0.onnx \
    --input_shape [1,3,32,32] \
    --mean_values [128.0,128.0,128.0] \
    --scale_values [255,255,255] \
    --model_name open_closed_eye

Demo

python3 demo.py \
    --model open_closed_eye.xml \
    --data_root ../../data/open_closed_eye/val/