Deep Convolutional Neural Networks for Kinship Prediction: An Effective Method

Main Article Content

M. H. Lohgaonkar

Abstract

Objective: In this paper, we worked on kinship verification, has many applications, such as, finding missing children, identify family and non-family member etc. our aim is to kinship prediction using similarity computation to identify kin and non-kin based on image dataset.


Method: To measure the accuracy of the proposed method on primary 96-family dataset includes 410 images and 77,887 different pairs. The data was split into 80% for training and 20% for testing. We proposed Siamese Deep Convolutional Neural Network model with deep algorithm.


Findings: We find that our suggested model performs better, with an average similarity score of 72.73%.


Novelty: The outcomes from our primary kinship datasets indicated that the proposed techniques outperformed both state-of-the-art kinship verification methods and human capabilities in our kinship verification task. The experimental results demonstrated the superior efficacy of our approach in comparison to existing methods and human performance.

Article Details

Section
Articles