terryoo / convolutional-network

ICIP 2017 Paper: Convolutional Neural Networks and Training Strategies for Skin Detection

Convolutional Neural Networks and Training Strategies for Skin Detection (ICIP 2017)

Abstract

This paper presents two convolutional neural networks (CNN) and their training strategies for skin detection. The first CNN consists of 20 convolution layers with 3 x 3 filters which is a kind of VGG network, and the second is composed of 20 network-in-network (NiN) layers, which can be considered a modification of Inception structure. When training these networks for human skin detection, we consider patch-based and whole-image-based training. The first method focuses on local features such as skin color and texture, and the second on the human-related shape features as well as color and texture. Experiments show that the proposed CNNs yield better performance than the conventional methods and also than the existing deep-learning based method. Also, it is found that the NiN structure generally shows higher accuracy than the VGG-based structure. The experiments also show that the whole-image-based training that learns the shape features yields better accuracy than the patch-based learning that focuses on local color and texture only.

All Experimental results on ECU, Pratheepan, and HGR dataset

Proposed method

• Patch based VGG method
• Patch based NiN method
• Image based VGG method
• Image based NiN method

Results

• [link]

Quantitative and Subjective Results

• Comparison of PR and ROC curves

• Visual comparison with outher methods on the Pratheepan dataset

convolutional-network's Issues

could you please publish the trained model?

thx!