Over time, numerous studies have been conducted to read license plates and recognize license plates. However, it is noteworthy that these studies usually do not have the ability to learn complex structures in images with high accuracy. For this purpose, this paper uses the high capacities of deep neural networks to learn license plate identifiers. The proposed model in this paper includes two main steps: highlighting license plates and reading the ID. In the proposed model, the support vector machine (SVM) network is used to select the best range. After identifying the range of the license plate, its characters must be recognized. In this step, a Gated convolutional neural network ((GCNN)) will be used. The proposed model is evaluated on two datasets, FZU Cars and Stanford Cars, and the results of the experiments show that this model has higher accuracy than other methods presented in both datasets.

Due to the ease of writing by hand and the inherent interest in it, writing by hand is still popular among many people. Considering the digitization of today's world and the massive amount of current information on paper, there is a need for a system to convert handwriting into its digital form to speed up access to information and reduce storage space. According to the research carried out in this field, recognizing Persian handwritten texts remains a relatively difficult issue due to the complex and irregular nature of writing and the diversity of people's handwriting. This research introduces a novel method to recognize handwritten texts at the sentence level. To use word recognition methods in sentence recognition, segmentation techniques are needed to separate the words in the sentence. The segmentation algorithm in handwritten texts is inefficient due to overlapping words. Since Recurrent neural networks (RNN) were a turning point in the recognition of correct writing, in this article, by removing the segmentation step, a new architecture, an RNN combined with a Gated Multi-scale convolutional neural network (GMCNN), is introduced in order to recognize handwritten sentences. Using the proposed architecture, recognizing Persian handwritten sentences in the Sadri dataset has a character error rate of 2.99%, a word error rate of 6.67%, and a sentence error rate of 36.87%. For further evaluation, the proposed method was also evaluated on IAM and Washington datasets. The results show that the proposed method outperforms other known algorithms.

Deep fake technology provides the possibility of automatic production and creation of (fake) video content through adversarial networks. Deep fake technology is a very challenging technology with very wide dimensions in various issues that have a deep impact on society, for example, this technology may cause bias in elections, create immoral content for the purpose of extortion, create political trends, etc. Much research has been devoted to the development of detection methods to reduce the potential negative impact of deep fake. The use of neural networks and deep learning is a common approach among all the research done in this field. In this research, the combination of GRU and LSTM neural networks has been used to detect fake. First, the Resnet neural network was used to extract the features of each frame, and then the GRU and LSTM layers were trained using these features, and finally, the fully connected layer was trained for the purpose of classification. The classification of input samples is used. To train and evaluate the proposed method, the training and evaluation datasets of FaceForensics++, Celeb-DF and Deepfake Detection Challenge have been used. The results show that the proposed method has obtained 97, 93 and 84% accuracy of diagnosis for the mentioned data sets and these results have improved by more than 2% compared to the reference research.

Deep convolutional neural networks (CNNs) have attained remarkable success in numerous visual recognition tasks. There are two challenges when adopting CNNs in real-world applications: a) Existing CNNs are computationally expensive and memory intensive, impeding their use in edge computing; b) there is no standard methodology for designing the CNN architecture for the intended problem. network pruning/compression has emerged as a research direction to address the first challenge, and it has proven to moderate CNN computational load successfully. For the second challenge, various evolutionary algorithms have been proposed thus far. The algorithm proposed in this paper can be viewed as a solution to both challenges. Instead of using constant predefined criteria to evaluate the filters of CNN layers, the proposed algorithm establishes evaluation criteria in online manner during network training based on the combination of each filter’s profit in its layer and the next layer. In addition, the novel method suggested that it inserts new filters into the CNN layers. The proposed algorithm is not simply a pruning strategy but determines the optimal number of filters. Training on multiple CNN architectures allows us to demonstrate the efficacy of our approach empirically. Compared to current pruning algorithms, our algorithm yields a network with a remarkable prune ratio and accuracy. Despite the relatively high computational cost of an epoch in the proposed algorithm in pruning, altogether it achieves the resultant network faster than other algorithms.

Although, speech recognition systems are widely used and their accuracies are continuously increased, there is a considerable performance gap between their accuracies and human recognition ability. This is partially due to high speaker variations in speech signal. Deep neural networks are among the best tools for acoustic modeling. Recently, using hybrid deep neural network and hidden Markov model (HMM) leads to considerable performance achievement in speech recognition problem because deep networks model complex correlations between features. The main aim of this paper is to achieve a better acoustic modeling by changing the structure of deep convolutional neural network (CNN) in order to adapt speaking variations. In this way, existing models and corresponding inference task have been improved and extended. Here, we propose adaptive windows convolutional neural network (AWCNN) to analyze joint temporal-spectral features variation. AWCNN changes the structure of CNN and estimates the probabilities of HMM states. We propose adaptive windows convolutional neural network in order to make the model more robust against the speech signal variations for both single speaker and among various speakers. This model can better model speech signals. The AWCNN method applies to the speech spectrogram and models time-frequency varieties. This network handles speaker feature variations, speech signal varieties, and variations in phone duration. The obtained results and analysis on FARSDAT and TIMIT datasets show that, for phone recognition task, the proposed structure achieves 1. 2%, 1. 1% absolute error reduction with respect to CNN models respectively, which is a considerable improvement in this problem. Based on the results obtained by the conducted experiments, we conclude that the use of speaker information is very beneficial for recognition accuracy.

Speaker recognition is a process of recognizing persons based on their voice which is widely used in many applications. Although many researches have been performed in this domain, there are some challenges that have not been addressed yet. In this research, Neutrosophic (NS) theory and convolutional neural networks (CNN) are used to improve the accuracy of speaker recognition systems. To do this, at first, the spectrogram of the signal is created from the speech signal and then transferred to the NS domain. In the next step, the alpha correction operator is applied repeatedly until reaching constant entropy in subsequent iterations. Finally, a convolutional neural networks architecture is proposed to classify spectrograms in the NS domain. Two datasets TIMIT and Aurora2 are used to evaluate the effectiveness of the proposed method. The precision of the proposed method on two datasets TIMIT and Aurora2 are 93.79% and 95.24%, respectively, demonstrating that the proposed model outperforms competitive models.