A Convolutional Neural Network Cascade for Face Detection
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
12net.lua fix 1 May 29, 2016
24net.lua fix 1 May 29, 2016
README.md README May 29, 2016
detector.lua

README.md

#A Convolutional Neural Network Cascade for Face Detection

Following this paper:

http://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Li_A_Convolutional_Neural_2015_CVPR_paper.pdf

This is an implementaton of a fast face detector based on my blog: https://deeplearningmania.quora.com/ An inspiration has been taken from the following paper: http://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Li_A_Convolutional_Neural_2015_CVPR_paper.pdf

Dependencies

This code is written in Torch7. To use it you will need:

A recent version of Torch7: https://github.com/torch/torch7/wiki/Cheatsheet#installing-and-running-torch To work with a GPU use CudaTensor: https://github.com/torch/cutorch

Data

FDDB: http://vis-www.cs.umass.edu/fddb/ AFLW: https://lrs.icg.tugraz.at/download.php PASCAL: http://host.robots.ox.ac.uk/pascal/VOC/databases.html

#IMPORTANT

This code contains only 12-net and 24-net convolutional networks. In order for the detector to give a high recall as stated in the blog - add a 48-net network to the pipeline.