A multi-color based features from facial images for automatic ethnicity identification model

Image mining is the method of searching and discovering valuable information and knowledge from a huge image dataset. Image mining is based on data mining, digital image processing, machine learning, image retrieval, and artificial intelligence. Image mining handled with the hidden information extraction, an association of image data and additional pattern which are not clearly visible in the image. Choosing the proper objects or the feature of the image to be suitable for image mining process is the main challenge would face the programmer. The process includes fine out the most efficient routes at a shorter time and saving the users effort. The main objective of this paper is to design and implement the image classification system with a higher performance, where a CIFAR-10 data set is used to train and testing classification models using CNN. A convolutional neural network is trustworthy, and it could lead to high-quality results. The high accuracy of 98% has been obtained using deep convolutional neural network (DCNN).

Section: Materials and Methods 21 Convolutional Neural Network (Cnn)mentioning

confidence: 99%

Deep image mining for convolution neural network

Jasm

Hamad

Alrawi

2020

“…Earlier layers of this CNN found a generic property. The layers of CNN [13], [14], on the other hand, become increasingly particular to the specifics of classes in the actual or real dataset. As a result, previous layers will aid in the extraction of current data features.…”

Section: Transfer Learning Techniquementioning

confidence: 99%

Novel deep learning model for vehicle and pothole detection

Gayathri

Thangavelu

2021

The most important aspect of automatic driving and traffic surveillance is vehicle detection. In addition, poor road conditions caused by potholes are the cause of traffic accidents and vehicle damage. The proposed work uses deep learning models. The proposed method can detect vehicles and potholes using images. The faster region-based convolutional neural network (CNN) and the inception network V2 model are used to implement the model. The proposed work compares the performance, accuracy numbers, detection time, and advantages and disadvantages of the faster region-based convolution neural network (Faster R-CNN) with single shot detector (SSD) and you only look once (YOLO) algorithms. The proposed method shows good progress than the existing methods such as SSD and YOLO. The measure of performance evaluation is Accuracy. The proposed method shows an improvement of 5% once compared with the previous methods such as SSD and YOLO.

“…To achieve this goal, six classifiers are being applied and their performances are compared to distinguish the most accurate one. Three fundamental stages of classification, face extraction, and feature extraction are suggested through Face Expression Recognition [11,12]. As a coral phase, feature selection is done to elecit the most important features that would have the most significance impact value [13] in which factors such as computational performance and the classification competence are manipulated by [14,15].…”

Section: Introductionmentioning

confidence: 99%

Classification techniques’ performance evaluation for facial expression recognition

Mahmood

Abdulrazaq

Zeebaree

et al. 2020

Facial exprestion recognition as a recently developed method in computer vision is founded upon the idea of analazing the facial changes in which are witnessed due to emotional impacts on an individual. This paper provides a performance evaluation of a set of supervised classifiers used for facial expression recognition based on minimum features selected by chi-square. These features are the most iconic and influential ones that have tangible value for result dermination. The highest ranked six features are applied on six classifiers including multi-layer preceptron, support vector machine, decision tree, random forest, radial baised function, and k-nearest neioughbor to figure out the most accurate one when the minum number of features are utilized. This is done via analyzing and appraising the classifiers’ performance. CK+ is used as the research’s dataset. Random forest with the total accuracy ratio of 94.23 % is illustrated as the most accurate classifier amongst the rest.