Intelligent bite marking analysis and classification using deep convolutional neural network based Xception model

Abstract

Bite mark analysis and classification play a vital role in forensics. The recent advances in computer vision and deep learning models paves a way for the design of automated bite mark detection and classification process. This article focuses on the design of intelligent bite marking analysis and classification using deep convolutional neural network based Xception model. The major goal of the proposed model is to determine the appropriate class labels for the bite marked images. The proposed model initially intends to pre-process the bite marked images in different ways such as hair removal, median filtering based noise removal, and adaptive histogram based contrast enhancement. Besides, Chan Vese Segmentation approach is applied for segmenting the bite marked images. The data augmentation process is performed for increasing the count of images. In addition, Xception model is employed for the extraction of features. Finally, two machine learning (ML) classifications such as support vector machine (SVM) and logistic regression (LR) models are employed for image classification. For demonstrating the enhanced performance of the presented models, a set of simulations were carried out on their own dataset and the results ensured the betterment of the proposed model over the other existing models.