Multi-Modal Region Based Convolution Neural Network (MM-RCNN) for Ethnicity Identification and Classification
C. Christy1, S. Arivalagan2, P. Sudhakar3

1C. Christy, Research Scholar, Department of CSE, Annamalai University 
2Dr. S. Arivalagan, Assistant Professor, Department of CSE, Annamalai University
3Dr. P. Sudhakar, Assistant Professor, Department of CSE, Annamalai University
Manuscript received on 28 August 2019. | Revised Manuscript received on 08 September 2019. | Manuscript published on 30 September 2019. | PP: 1168-1176 | Volume-8 Issue-11, September 2019. | Retrieval Number: J91020881019/2019©BEIESP | DOI: 10.35940/ijitee.J9102.0981119
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Human facial images help to acquire the demographic information of the person like ethnicity and gender. At the same time, the ethnicity and gender acts as a significant part in the face-related applications. In this study, image-based ethnicity identification problem is considered as a classification problem and is solved by deep learning techniques. In this paper, a new multi-modal region based convolutional neural network (MM-RCNN) is proposed for the detection and classification of Ethnicity to determine the age, gender, emotion, ethnicity and so on. The presented model involves two stages namely feature extraction and classification. In the first stage, an efficient feature extraction model called ImageAnnot is developed for extracting the useful features from an image. In the second stage, MM-RCNN is employed to identify and then classify ethnicity. To validate the effective performance of the applied MM-RCNN model, various evaluation parameters has been presented and the simulation outcome verified the superior nature of the presented model compared to existing models.
Keywords: Classification, Deep learning, Ethnicity, Faster R-CNN, Feature extraction
Scope of the Article: