Sharif Digital Repository

In the extensive horizon of artificial intelligence technology, one of the grand challenges in computer vision has been semantic segmentation. This task which aimed to predict label for each pixel of image, describes the scene, due to the need of low level information, is more complicated in comparison with other computer vision tasks. However, as part of concept of scene understanding and a crucial step in many real world applications such as autonomous driving, human-computer interaction and robot navigation, many researchers have been sought to resolve it. What makes this task more challenging rather than other computer vision tasks is that information beyond a pixel, its neighbors and... 

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... 

Semantic segmentation is the tash of labeling each pixel of an input image. It is one of the main problems in computer vision and plays an important role in scene understanding. State of the art methods of solving it are based on Convolutional Neural Networhs (CNNs). While many real world tasks like autodriving cars and robot navigation require fast and lightweight models, CNNs inherently tend to give beter accuracy when they are deeper and bigger, and this has raised interest in designing compact networks. Knowledge distillation is one of the popular methods of training compact networhs and helps to transfer a big and powerful network’s knowledge to a small and compact one. In this research... 

The labeling process is one of the most important tasks in the field of computer vision. The dense labeling problem is the main step towards 2D and 3D scene understanding. The main goal of dense labeling is to label all pixels of images that are known as a semantic segmentation of images in the related literature. Although the state-of-the-art results are mainly achieved by deep learning methods, traditional methods had also been at the center of attention for some years. In the last decades, convolutional neural networks have changed the landscape of visual recognition tasks such as labeling and semantic segmentation. The most important issues in deep learning models are the hardware and... 

Semantic segmentation which is the classification of every pixel in an input image is a fundamental task in the fields of computer vision and scene understanding. Applications of semantic segmentation include usage in autonomous vehicles and robotics. Since in this task dense annotation of images in the dataset is needed, recent methods have been proposed to utilize weakly-supervised and semi-supervised learning using data with weak labels and unlabeled data respectively. Because the amount of fully labeled data might not be sufficient in such methods, some papers have proposed to employ depth input data due to its rich geometrical and local information when available. In this research, an... 

Despite recent advancements in deep learning methods, these methods rely on a huge amount of training data to work. Recently the problem of solving classification and recently semantic segmentation problems with a few training data have gained attention to tackle this issue. In this research, we propose a meta-learning method by combining optimization-based and prototypical approaches in which a small portion of parameters are optimized with task-specific initialization. In addition to this and designing other parts of the method, we propose a new approach to use query data as an unlabeled sample to enhance task-specific learning. Alongside the mentioned method, we propose an approach to use... 

Despite widespread applications and high performance of deep neural networks in the fields of computer vision, they have been shown to be vulnerable to adversarial examples. An adversarial example is a perturbated image that the magnitude of its difference with its corresponding natural image is small and yet given such example, the network produces incorrect output. In recent years, many approaches have been proposed to increase the robustness of DNNs against adversarial examples with adversarial training being proposed as the most effective defense measure. Approaches based on adversarial training try to increase the robustness of the network by training on the adversarial examples. One of... 

One of the most common forms of three-dimensional data is point clouds. In addition to its high flexibility in storing three-dimensional space, this type of data is the closest type of data to the output of three-dimensional sensors. Semantic segmentation of point clouds is a fundamental operation on this type of data, with applications in robotics, self-driving cars, virtual reality, remote sensing, and other fields that work with this type of data. Since deep learning models require abundant data for training, this type of data is not an exception to this rule with these models. However, the problem is that collecting and labeling this type of data is more difficult and costly compared to... 

Image segmentation in medical imaging, as one of the most important branches of medical image analysis, often faces the challenge of limited labeled data for application in deep learning methods. The high cost of data collection and the need for expertise in image segmentation, particularly in three-dimensional images such as MRI and CT or sequence images like CMR, have all contributed to this problem, even for popular networks like U-Net, which struggle to achieve high accuracy. As a result, research efforts have focused on semi-supervised learning approaches, weakly supervised learning, as well as multi-instance learning in medical image segmentation. Unfortunately, each of these methods... 

Wall detection, as a part of scene understanding and indoor 3D modeling, can be used in robotic, architecture, and augmented reality. In robotic, scene analysis and understanding is knows as one of the main steps in robot navigation and simultaneous localization and mapping and walls need to be detected to form a 3D map of indoor environment. Ceiling and floor are also important elements of an indoor environment, therefore we can see our problem as a room layout estimation if we consider ceiling and floor. Using a deep learning structure, we estimate room layout in a semantic segmentation manner. Our approach is real-time. Our proposed method uses a deep fully-convolutional network, layout... 

Crack detection is a critical task in monitoring and inspection of civil engineering structures. This study proposes a deep-learning-based model for automatic crack detection on the concrete surface. The proposed model is an encoder-decoder model which uses EfficientNetB7, the state-of-the-art convolutional neural network, as encoder and the U Net’s expansion path as decoder. To minimize the training time and maximizing the accuracy, we use transfer learning in our approach. We train our model with a novel training strategy on images from an open-source dataset and achieve 96.44% F1-score for unseen test data. To compare the performance of the proposed method, we evaluate our model on CFD... 

In image processing applications, sensors (Camera, LiDAR and Stereo) are essential for scene perception and Deep learning methods outperform most of the image processing tasks like 3D and 2D object detection and semantic segmentation. Different sensors are used in image processing tasks. Sensor fusion is using multiple sensors data to get better performance. Each sensor captures different data (e.g, color, texture, and depth). Some of them are distorted in inclement weather, intense illuminance changes, and dark environments which multi-sensor data fusion is used to overcome sensor weaknesses. One of the most important fields that sensor fusion used is Auto Driving cars (AD). Different... 

The health of the pavement ensures the safety and convenience of drivers and passengers. In the past few decades, pavement management systems have encountered challenges that often have produced solutions with excessive demand for resources, but low-accuracy results. New approaches must be developed in order to quickly and economically identify pavement failure, especially cracks. In recent years, researchers have focused on identifying pavement failures, but previous methods only worked on images that solely included pavements and cracks. However, when foreign objects such as cars and vegetation were present, these methods were not as effective. To improve upon these methods, semantic... 

Most of the existing visual SLAM methods heavily rely on a static world assumption and easily fail in dynamic environments. One solution is to eliminate the influence of dynamic objects by introducing deep learning-based semantic information to SLAM systems. In this project, we propose a real-time semantic RGB-D SLAM (built upon RTAB-Map) system for dynamic environments that is capable of detecting moving objects and maintaining a static map for robust camera tracking. Furthermore, we augment the semantic segmentation process using an Extended Kalman filter module to detect temporarily static moving objects by adding centroids to each found dynamic object and calculating their velocity. We... 

Event cameras are bio-inspired sensors. They have outstanding properties compared to frame based cameras: high dynamic range (120 vs 60), low latency, no motion blur. Event cameras are appropriate for using in challenging scenarios such as vision system in self-driving cars and they have been used for high level machine vision tasks such as semantic segmentation, depth estimation. In this project, we worked on semantic segmentation using an event camera for self-driving cars. i) We introduce a large dataset, our dataset was produced using Carla simulator and it contains RGB images, events and semantic segmentation labels. ii) This project introduces new event based semantic segmentation...