Sharif Digital Repository

Recent approaches in the field of semantic image segmentation rely on deep networks that are trained by pixel-level labels. This level of labeling requires a lot of time for the labeler person; because these networks require large training datasets to achieve optimal accuracy and the lack of data at the labeled pixel level causes a significant drop in their performance. In order to overcome this problem, weakly supervised segmentation approaches have been proposed. In these approaches, weaker labels such as image-level labels, bounding boxes, scribbles, etc. have been introduced to train the networks.In this thesis, a method for segmentation of kidney and kidney tumor in CT scan images based... 

The human action recognition is one of the most important concepts of computer vision in recent decades. Most of the two dimensional methods in this field are facing serious challenges such as occlusion and missing the third dimension of data. Development of depth sensors has made easy access to tracking people and 3D positions of human body joints. This Thesis proposes a new method of action recognition that utilizes the position of joints obtained by Kinect sensor. The learning stage uses Fisher Linear Discriminant Analysis (LDA) to construct discriminant feature space. Two types of distances, i.e., Euclidean and Mahalanobis, are used for recognizing the states. Also, Hidden Markov Model... 

The non-stationary nature of brain activity signals and their many inter-subject differences have created many challenges in the practical applications of emotion recognition based on electroencephalogram (EEG) signals, such as brain-computer interfaces. In such a way, the use of traditional classifiers in classifying these signals leads to a significant decrease in accuracy when applying the classifier to a new subject. Domain Adaptation methods seem to be an effective way to solve this problem by minimizing the difference between the EEG signals of different subjects. But in the basic techniques for domain adaptation, looking at all subjects' data in the same look causes the loss of a part... 

In this project, a novel action recognition method, named Fisherposes, is proposed, which is improved by several space-time (spatio-temporal) methods afterwards. The proposed method utilizes skeleton data obtained from Kinect sensor. First, pre-processing is performed in which the scales of bodies are canceled and the skeletons become aligned in order to make the method robust to location, orientation, and scale of people. In Fisherposes method, every action is defined as a sequence of body poses. Using the training samples for the poses, a Fisher subspace is created which we name it Fisherposes. Moreover, a novel distance measuring function, named regularized Mahalanobis distance, is... 

Continuous Blood Pressure monitoring can provide invaluable information about individuals’ health conditions. However, BP is conventionally measured using inconvenient cuff-based instruments, which prevents continuous BP monitoring. This work presents an efficient algorithm, based on the Pulse Arrival Time (PAT), for the continuous and cuff-less estimation of the Systolic Blood Pressure (SBP), Diastolic Blood Pressure (DBP), and Mean Arterial Pressure (MAP) values. The proposed framework estimates the BP values through processing vital signals and extracting two types of features, which are based on either physiological parameters or whole-based representation of vital signals. Finally, the... 

Brain states analysis during consciousness is emerging research in brain-computer interface(BCI). Emotion recognition can be applied to learn brain states and stages of neural activities. Therefore, emotion recognition is crucial to the analysis of brain functioning. Electrical signals such as electroencephalogram (EEG), electrocardiogram (ECG) and functional magnetic resonance imaging(fMRI) are frequently used in emotion recognition researches. Convenience in recording, non-invasive nature and high temporal resolution are the factors that have made EEG popular in brain researches. EEG can be used to identify brain region activity solely or the connectivity of various regions in time in the... 

In Iranian Sign Language (ISL), alongside the movement of fingers, the movement of the lips is also essential for to perform words completely and correctly. The purpose of current study is to provide an automated lip-reading system using deep neural networks and implement it on Rasa social robot; So that the robot can recognize a limited number of specified Persian words. To do this, we propose an automated lip-reading system based on convolutional neural networks and long short-term memories. Convolutional neural networks in extracting features from images and long short-term memories in modeling temporal dynamics have achieved good results. We have also recorded a database in Persian... 

Action recognition has a lot of applications in everyday human life. In the past, the researchers concentrated on using RGB frames, but since the advent of 3-dimensional sensors such as Kinect, 3D action recognition drew researchers' attention. Kinect can extract the joints of the body in action as time series. One of the main challenges of action recognition is that different individuals perform an action with various styles and speeds. Hence, the conventional methods such as calculating Euclidean distance seem inappropriate for this task. One solution is to use the techniques such as DTW, which aims to temporal aligning of the sequences. The DTW is not a metric distance; hence, in this... 

Analysis of people appearance variation in multi-camera networks for person re-identification or person retrieval is a very challenging problem due to the many intra-class variations between different cameras. Like any problem in the field of machine vision, it is generally divided into two parts. The first part is feature extraction and the second part is feature matching for person retrieval. So far, various methods have been proposed for the extraction of discriminative features, which are generally divided into three categories: stripe-based, patch-based, and body-based methods. However, methods based on stripes, although simpler, have performed better due to their greater compatibility... 

Object tracking is one of the first, most basic and among the topics of interest in the field of computer vision. Nowadays, with the availability of high-quality and inexpensive video cameras and the expansion of neural networks, there has been a great interest in automatic video analysis using object tracking algorithms. However, many of the existing object tracking algorithms do frame-by-frame tracking using videos with high frame rates, which is not suitable for all locations that use surveillance cameras, because due to existing hardware limitations, the recorded videos are either kept for a limited period of time or are forcibly stored with low frame rates, which leads to the loss of a... 

This thesis suggests a practical system designed for the implementation of vision-based virtual assistants, aligning with the identified needs in the research background. Distinguishing itself from current literature, this study thoroughly explores and delineates the computing hierarchy while also suggesting the appropriate software architecture customized for the effective utilization of virtual assistants. Focusing on the most common application in vision-based virtual assistants—object detection and tracking—the research introduces an efficient multi-object tracking method using the tracking-by-detection paradigm. Notably, the proposed tracker stands out for its minimal computational... 

Epilepsy is a chronic neurological disorder characterized by recurrent and abrupt seizures. Seizures occur due to disturbances in the interactions between the distributed neuronal populations in the brain. Investigation of the brain functional connectivity networks is a way to better understand how the brain functions during seizure. To estimate the brain functional connectivity network, we need criteria that can estimate the functional connections between the brain regions from the recorded brain functional data such as electroencephalogram (EEG) signals. After estimating the functional brain connectivity networks, it is possible to create graphs corresponding to these estimated networks... 

In recent years, the simultaneous analysis of image and text by artificial intelligence has gained much attention. Video description is one of the topics used to help the blind, automate video content analysis, and more. This issue is usually examined in the context of supervised learning and is very similar to image description and machine translation.The proposed solutions to this problem are mainly in the framework of encoder-decoder and attention-based neural networks. Selection of various pre-trained networks to extract 2D and 3D visual features (description of objects and actions in the image), various hierarchical structures and different teaching methods (based on reinforcement... 

T lymphocytes or T cells are a type of white blood cell that play an important role in the immune system. Memory T cells are a subset of them that are able to reactivate when being re-exposed to the pathogen. Because of the properties of these cells, they are an attractive choice for immunotherapy. Initial memory subgroups were shown to be more persistently effective in immunotherapies. However little is known about development and differentiation of these subgroups. With the discovery of a new subset of T cells, called T memory stem cells (TSCM), They are considered to develop through four stages, naive T cells (TN), T memory stem cells (TSCM), central memory T cells (TCM) and effector... 

The most important issues in massive concrete structures such as concrete dams are the increase in temperature due to heat generated by the dam Hydration of cement in concrete. Internal restrains are created from thermal gradient between the surface and the interior of the structure. External restrains are created from connections between structures and foundation for obstruct the movement of concrete structures due to temperature change. External restrains and internal creates tensile stresses in the concrete considerable, which can lead to thermal cracks in the structure. To prevent or minimize the possibility of cracks are examined thermal studies for the design of reinforced concrete... 

Coding information data is one of the ways to prevent noise effect on the information bits, during passing communication channels. Coding data gives the possibility to detect and correct the data in the receiver. The LDPC codes which were first introduced by Gallager in the 1962 are forward error correction codes and can approach the Shannon’s capacity to within hundredths of a decibel.
In this project a modified algorithm for decoding these group of codes is introduced, which can achieve acceptable bit error rate, while it can have better throughput than the same implementations. This algorithm is implemented partial-parallel for IEEE802.11n standard in ASIC. It is shown that, it has...