Abstract:This work showed the capability of handling large number of classes for classification with human cognition inspired methods. A cognition based techniques for both feature extraction, (self-similarity feature, Intensity Level Multi Fractal Dimension (ILMFD)) as well as classification purpose (decision tree clustering based multi-level Artificial Neural Network classifier-MLANN-DTC) were employed to implement facial recognition based object detection system. A DTC based approach reduces the search space time and also provides opportunity for very less amount of classes (a smaller part of the large number of classes) to be handled by the respective classifier for classification. It also mimics fast recognition capability of humans. In this work, two different databases were used for experiment, first one is our own collected facial images from rotation based video clips (117 persons and 40 facial images per person) named as NS database, and other is standard ORL database (40 persons and 10 facial images per person). In pre-processing step, the facial images were segmented to obtain facial part using context window based texture of pixels (CWTP) & back-propagation neural network (BPNN) based model and then a scale and rotation independent ILMFD feature was computed from each segmented image. Further, a combination of K-means and hierarchal clustering was used to build super classes. All classes' data were distributed among these 6 super classes (heuristically chosen) for own NS database and 3 for ORL database as per their similarity based on ILMFD features. Multi-level ANNs models were employed for all super classes and further their classification results were fed into decision clustering based model to obtain fine-tuned results, which showed significant improvement in terms of classification efficiency. This approach believes in center tendency of largest cluster to refer the actual class decision from multiple decisions obtain corresponding to multiple input data of the same class. In this work, the MLANN-DTC based proposed model has produced 89.542 ± 1.167% and 87.098 ± 2.066% classification efficiency (± standard deviation) for single input and for group based decision (decision clustering), 95.042 ± 0.719% and 89 ± 2.549% for NS and ORL database respectively. This improved classification results motivate its application for other object recognition and classification problems. The basic idea of this work also supports better handling of classification which deals with a large number of classes.

Object recognition using cognition based decision tree clustering in multi-level artificial neural network classifier and self similarity as feature criterion

Significant Enhancement of Object Recognition Efficiency Using Human Cognition based Decision Clustering

MTCNN and FACENET Based Access Control System for Face Detection and Recognition

Efficient Object Segmentation and Recognition Using Multi-Layer Perceptron Networks

Intelligent Education Managemnt System Design for Universities based on MTCNN Face Recognition Algorithm

IntelligentFaceNet: Designing a Multi-Cascaded Attentive and Adaptive Deep Learning Network for Facial Recognition using Heuristic Approach

Improved multiview biometric object detection for anti spoofing frauds

Classification of fused face images using multilayer perceptron neural network

Face recognition with illumination, scale and rotation invariance using multiblock LTP-GLCM descriptor and adaptive ANN

An efficient facial emotion recognition system using novel deep learning neural network-regression activation classifier

Detection, Segmentation and Recognition of Face and its Features Using Neural Network

Facial Recognition Using Convolutional Neural Network Using Real-Time Data

Expression invariant face recognition based on multi-level feature fusion and transfer learning technique

New Face Recognition System Based on DCT Pyramid and Backpropagation Neural Network

Emotion Recognition for Challenged People Facial Appearance in Social using Neural Network

A hybrid approach for face recognition using a convolutional neural network combined with feature extraction techniques

Identifying People's Faces in Smart Banking Systems Using Artificial Neural Networks

Classification of Fused Images using Radial Basis Function Neural Network for Human Face Recognition

A face recognition system using convolutional feature extraction with linear collaborative discriminant regression classification

Multi-Object Classification and Unsupervised Scene Understanding Using Deep Learning Features and Latent Tree Probabilistic Models

Deep Residual Adaptive Neural Network Based Feature Extraction for Cognitive Computing with Multimodal Sentiment Sensing and Emotion Recognition Process