Artificial Intelligence
S. H. Zahiri; R. Iranpoor; N. Mehrshad
Abstract
Background and Objectives: Person re-identification is an important application in computer vision, enabling the recognition of individuals across non-overlapping camera views. However, the large number of pedestrians with varying appearances, poses, and environmental conditions makes this task particularly ...
Read More
Background and Objectives: Person re-identification is an important application in computer vision, enabling the recognition of individuals across non-overlapping camera views. However, the large number of pedestrians with varying appearances, poses, and environmental conditions makes this task particularly challenging. To address these challenges, various learning approaches have been employed. Achieving a balance between speed and accuracy is a key focus of this research. Recently introduced transformer-based models have made significant strides in machine vision, though they have limitations in terms of time and input data. This research aims to balance these models by reducing the input information, focusing attention solely on features extracted from a convolutional neural network model. Methods: This research integrates convolutional neural network (CNN) and Transformer architectures. A CNN extracts important features of a person in an image, and these features are then processed by the attention mechanism in a Transformer model. The primary objective of this work is to enhance computational speed and accuracy in Transformer architectures. Results: The results obtained demonstrate an improvement in the performance of the architectures under consistent conditions. In summary, for the Market-1501 dataset, the mAP metric increased from approximately 30% in the downsized Transformer model to around 74% after applying the desired modifications. Similarly, the Rank-1 metric improved from 48% to approximately 89%.Conclusion: Indeed, although it still has limitations compared to larger Transformer models, the downsized Transformer architecture has proven to be much more computationally efficient. Applying similar modifications to larger models could also yield positive effects. Balancing computational costs while improving detection accuracy remains a relative goal, dependent on specific domains and priorities. Choosing the appropriate method may emphasize one aspect over another.
Image Processing
A. Ghanbari Talouki; A. Koochari; S. A. Edalatpanah
Abstract
Background and Objectives: Images and videos play significant roles in our lives. Moreover, there are considerable improvements in computer data collection systems; therefore, anyone is able to acquire much many images or videos, whereas he/she cannot process them manually. Images and videos became attractive ...
Read More
Background and Objectives: Images and videos play significant roles in our lives. Moreover, there are considerable improvements in computer data collection systems; therefore, anyone is able to acquire much many images or videos, whereas he/she cannot process them manually. Images and videos became attractive since depicting and digital processing of these kinds of data became possible. Since indeterminacy surrounded the world, including images and videos of that; imprecision is needed to interpret this world.Methods: Neutosophic logic, which is from philosophy and is also included of logic, set theory and probability/statistics, is able to depict this imprecision. As a result, advanced image processing can be defined by translation of image processing into neutrosophic domain. In this paper, first of all, a general introduction about image/video processing (segmentation, noise reduction and image retrieval) and uncertainty is stated. Then, definitions of fuzzy sets, intuitionistic fuzzy sets and neutrosophic sets are expressed. In the following, applications of neutrosophic domain in image and video processing such as segmentation, noise reduction and image retrieval are introduced.Results: Although, neutrosophic is used for image restoration and segmentation; input images are usually medical images or gray level natural images. The remarkable point is that there are few researches that focuse on image restoration or segmentation using neutrosophic and consider color images. Therefore, color image restoration and segmentation in neutrosophic environment is novel to be done.Conclusion: Neutrosophic usage in image retrieval brings out an improvement in average recall and precision measure compared to earlier methods. Considering different textures and shapes can extend usages of neutrosophic space in image retrieval applications.
Computer Vision
M. Taheri; M. Rastgarpour; A. Koochari
Abstract
Background and Objectives: medical image Segmentation is a challenging task due to low contrast between Region of Interest and other textures, hair artifacts in dermoscopic medical images, illumination variations in images like Chest-Xray and various imaging acquisition conditions.Methods: In ...
Read More
Background and Objectives: medical image Segmentation is a challenging task due to low contrast between Region of Interest and other textures, hair artifacts in dermoscopic medical images, illumination variations in images like Chest-Xray and various imaging acquisition conditions.Methods: In this paper, we have utilized a novel method based on Convolutional Neural Networks (CNN) for medical image Segmentation and finally, compared our results with two famous architectures, include U-net and FCN neural networks. For loss functions, we have utilized both Jaccard distance and Binary-crossentropy and the optimization algorithm that has used in this method is SGD+Nestrov algorithm. In this method, we have used two preprocessing include resizing image’s dimensions for increasing the speed of our process and Image augmentation for improving the results of our network. Finally, we have implemented threshold technique as postprocessing on the outputs of neural network to improve the contrast of images. We have implemented our model on the famous publicly, PH2 Database, toward Melanoma lesion segmentation and chest Xray images because as we have mentioned, these two types of medical images contain hair artifacts and illumination variations and we are going to show the robustness of our method for segmenting these images and compare it with the other methods.Results: Experimental results showed that this method could outperformed two other famous architectures, include Unet and FCN convolutional neural networks. Additionally, we could improve the performance metrics that have used in dermoscopic and Chest-Xray segmentation which used before.Conclusion: In this work, we have proposed an encoder-decoder framework based on deep convolutional neural networks for medical image segmentation on dermoscopic and Chest-Xray medical images. Two techniques of image augmentation, image rotation and horizontal flipping on the training dataset are performed before feeding it to the network for training. The predictions produced from the model on test images were postprocessed using the threshold technique to remove the blurry boundaries around the predicted lesions.
F. Yaghmaee; M. Kamyar
Abstract
To distinguish between human user and computer program to enhance security, a popular test called CAPTCHA is used on Web. CAPTCHA has an important role in preventing Denial Of Service (DOS) attacks in computer networks. There are many different types of CAPTCHA in different languages. Due to the expansion ...
Read More
To distinguish between human user and computer program to enhance security, a popular test called CAPTCHA is used on Web. CAPTCHA has an important role in preventing Denial Of Service (DOS) attacks in computer networks. There are many different types of CAPTCHA in different languages. Due to the expansion of Persian-language and documents on internet, creating a suitable Persian CAPTCHA seems to be necessary. In this paper, we introduce three different types for Persian CAPTCHA in different domains. In the first type, based on the particular characteristics of Persian writing such as contiguous writing and image processing techniques, high strength CAPTCHA is provided. In the second type, the meaning of Persian words are used to creating CAPTCHA and in the third type, the combination of image processing techniques and the meaning of Persian words are used. Experimental results show that proposed CAPTCHAs has high security against attacks while Persian people can easily recognize them.
A. Safaei; M. Jahed
Abstract
Gesture and motion recognition are needed for a variety of applications. The use of human hand motions as a natural interface tool has motivated researchers to conduct research in the modeling, analysis and recognition of various hand movements. In particular, human-computer intelligent interaction has ...
Read More
Gesture and motion recognition are needed for a variety of applications. The use of human hand motions as a natural interface tool has motivated researchers to conduct research in the modeling, analysis and recognition of various hand movements. In particular, human-computer intelligent interaction has been a focus of research in vision-based gesture recognition. In this work, we introduce a 3-D hand model recognition method that offers flexible and elaborate representation of hand motion. We used landmark points on the tips and joints of the fingers and calculated the 3-D coordinates of these points through a stereo vision system followed by a Hidden Markov Model (HMM) to recognize hand motions. Experimentally, in an effort to evaluate the formation of hand gestures similar to those used in rehabilitation sessions, we studied three evolving motions. Given the natural hand features and uncontrolled environment, we were able to classify and differentiate unnatural slowness or rapidness in the performance of such motions, ranging from 45% to 93%.
