convolution neural network

In this study, the design of a reliable detection system that is able to identify different emotions with the desired accuracy has been considered. To reach this goal, two different structures for the emotion recognition system include 1) using linear and non-linear features of the electroencephalography (EEG) signal along with common classifiers and 2) using EEG signal in a deep learning structure is considered to identify emotional states. To design the system, the EEG signals of the DEAP database which were recorded by displaying emotional videos from 32 subjects were used. After the preparation and noise removal, linear and non-linear features such as: Skewness, Kurtosis, Hjorth parameters, Lyapunov exponent, Shannon entropy, correlation and fractal dimension and time reversibility were extracted from the alpha, beta and gamma subbands of the EEG signals. Then according to structure 1, the features were applied as input to common classifiers such as decision tree (DT), k nearest neighbor (kNN) and support vector machine (SVM). Also in structure 2, the EEG signal was considered as the input of the convoloutional neural network (CNN). The goal is to evaluate the results of deep learning networks and other methods for emotion recognition. According to the obtained results, the SVM achieved the best performance for identifying four emotional states with 94.1 % accuracy. Also, the proposed CNN identified the desired emotional states with the accuracy of 86%. Deep learning methods are superior to simple classifiers because they do not require the features of the signals and are resistant to different noises. Using a short period of time for the signals and performing near optimal preprocessing and conditioning, can further improve the results of deep neural networks.

Over the past few years, real-time classification of vehicle types has become an increasingly popular and important topic, given its wide range of applications in traffic control and analysis. Among the various methods available for classifying car types, convolutional neural networks (CNNs) have emerged as particularly appealing. In this article, we introduce a new real-time CNN architecture that is specifically designed to detect different types of cars. This innovative structure incorporates several unique features, including novel network architecture and structural elements, as well as an advanced learning method based on the back-propagation algorithm. One key aspect of our proposed method is its use of feature extraction at three different locations within the network, which allows for more accurate and efficient classification of car types. To evaluate the performance of our approach, we conducted experiments on two popular datasets (IRVD and MIO-TCD), and compared our results against those obtained using traditional CNN structures. Our evaluation results demonstrate that our proposed CNN architecture outperforms existing approaches, achieving superior classification accuracy across a range of criteria. Overall, our work represents a significant advance in the field of real-time car type classification, with broad implications for traffic management and analysis.

Sewage flow path is the main component of urban sewerage network infrastructure. Damage to sewers is less noticeable due to invisibility, and this failure to handle the damage leads to emergencies and unreasonable costs. These vital arteries need to be maintained and rebuilt during service for optimal performance in all dimensions. Nowadays, the methods of processing and classifying photos and videos taken by mobile videometer robots are widely used to inspect the sewer network. One of the successful algorithms in the field of image processing is the convolutional neural network algorithm, which is a subset of deep learning algorithm. In this paper, a convolutional neural network algorithm is used to classify images of sewer network damage and cases affecting the improvement, accuracy and performance of this algorithm. The images were obtained by a videometric robot from the sewer network. Results of using the proposed algorithm in the sewerage network, achieving 98% accuracy in classifying network faults and compared to other methods and also reducing the relatively low execution time of the proposed architecture (91 minutes) compared to other architectures valid ones are the same in deep learning on the same hardware platform. Also, in the future, the proposed algorithm will be used to analyze networks without the need for specialized personnel and also to control an automatic network videometry robot.