Skip to main content

RateMe is a neural network that allows you to recognize gestures of thumb up and thumb down

Project description

Rating system using computer vision

Introduction

RateMe is a neural network that allows you to recognize gestures of thumb up and thumb down. The algorithm can be embedded in your project and automate the process of evaluation of something or someone.

For example, using Rate Me you can:

  • Rate music on the YouTube Music service, when you are uncomfortable with clicking buttons

  • Count the votes in a beauty contest to determine the winner

  • Rate drinks and meals during their tasting

Technical part

Dependencies

opencv-contrib-python, numpy

matplotlib, jupyter for jupyter-notebook

pip3 install opencv-contrib-python numpy

Usage

Open Test_RateMe.ipynb to test RateMe on example images (for that you need to start jupyter-notebook, it will open window in your browser, where you need to select Test_RateMe.ipynb).

For example of realtime detection run detect_on_webcam.py.

Minimal working example:

import cv2
from RateMe.utils import RateMe

img = cv2.imread('test_imgs/like.jpg')
label = net.decode(img)

Description

RateMe is based on tiny-YOLOv3 architecture.

It's accuracy of thumb up/down gesture recognition is calculated as mean average precision (mAP@0.25) = 0.851941, or 85.19%; average IoU = 73.89%

The neural network has been trained on ~3K images (taken from different angles photos of people showing their thumbs or not). Images were labeled using the labelImg program.

Class labels: 0 -- "Like (thumb up)", 1 -- "Dislike (thumb down)"

Speed

Full pipeline speed is 6-7 FPS on Intel(R) Core(TM) i5-4300M CPU @ 2.60GHz.

~100ms on frame grabbing

~100ms on neural network inference

Project details


Release history Release notifications

This version

0.1.0

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for rateme, version 0.1.0
Filename, size File type Python version Upload date Hashes
Filename, size rateme-0.1.0-cp3-none-any.whl (32.2 MB) File type Wheel Python version cp3 Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page