Matting, which aims to separate the foreground object from the background of an image, is an important problem in computer vision. Most existing methods rely on auxiliary information such as trimaps or scibbles to alleviate the difficulty arising from the underdetermined nature of the matting problem. However, such methods tend to be sensitive to the quality of auxiliary information, and are unsuitable for real-time deployment. In this paper, we propose a novel Attention-based Multi-scale Matting Network (AMMNet), which can estimate the alpha matte from a given RGB image without resorting to any auxiliary information. The proposed AMMNet consists of three (sub-)networks: 1) a multi-scale neural network designed to provide the semantic information of the foreground object, 2) a Unet-like network for attention mask generation, and 3) a Convolutional Neural Network (CNN) customized to integrate high- and low-level features extracted by the first two (sub-)networks. The AMMNet is generic in nature and can be trained end-to-end in a straightforward manner. The experimental results indicate that the performance of AMMNet is competitive against the state-of-the-art matting methods, which either require additional side information or are tailored to images with a specific type of content (e.g., portrait). / Thesis / Master of Applied Science (MASc)

Functional near-infrared spectroscopy (fNIRS) is a neuroimaging modality that measures the hemodynamic responses to brain activation. With its cost-effectiveness and portability, fNIRS can be utilized to measure brain signals in the everyday world. However, factors such as blood pressure, cardiac rhythms, and motion can obscure the hemodynamic response function (HRF) obtained in fNIRS data. Motion, in particular, poses a significant challenge in obtaining the HRF for measurements conducted in everyday world activities when the subject is free to move. To address this, the General Linear Model (GLM) with temporally embedded Canonical Correlation Analysis (tCCA) has been shown to be effective in extracting the HRF by reducing motion and other systemic interferences. Recently, deep learning methods have also demonstrated its potential for time-series data analysis. The objective of this project is to evaluate the effectiveness of a novel transformer-based deep learning approach in comparison to the tradition method of GLM with tCCA

Interactive deep learning frameworks are crucial for effectively exploring and analyzing complex image datasets in visual analytics. However, existing approaches often face challenges related to inference accuracy and adaptability. To address these issues, we propose ImageSI, a framework integrating deep learning models with semantic interaction techniques for interactive image data analysis. Unlike traditional methods, ImageSI directly incorporates user feedback into the image model, updating underlying embeddings through customized loss functions, thereby enhancing the performance of dimension reduction tasks. We introduce three variations of ImageSI, ImageSI$_{text{MDS}^{-1}}$, prioritizing explicit pairwise relationships from user interaction, and ImageSI$_{text{DRTriplet}}$ and ImageSI$_{text{PHTriplet}}$, emphasizing clustering by defining groups of images based on user input. Through usage scenarios and quantitative analyses centered on algorithms, we demonstrate the superior performance of ImageSI$_{text{DRTriplet}}$ and ImageSI$_{text{MDS}^{-1}}$ in terms of inference accuracy and interaction efficiency. Moreover, ImageSI$_{text{PHTriplet}}$ shows competitive results. The baseline model, WMDS$^{-1}$, generally exhibits lower performance metrics. / Master of Science / Interactive deep learning frameworks are crucial for effectively exploring and analyzing complex image datasets in visual analytics. However, existing approaches often face challenges related to inference accuracy and adaptability. To address these issues, we propose ImageSI, a framework integrating deep learning models with semantic interaction techniques for interactive image data analysis. Unlike traditional methods, ImageSI directly incorporates user feedback into the image model, updating underlying embeddings through customized loss functions, thereby enhancing the performance of dimension reduction tasks. We introduce three variations of ImageSI, ImageSI$_{text{MDS}^{-1}}$, prioritizing explicit pairwise relationships from user interaction, and ImageSI$_{text{DRTriplet}}$ and ImageSI$_{text{PHTriplet}}$, emphasizing clustering by defining groups of images based on user input. Through usage scenarios and quantitative analyses centered on algorithms, we demonstrate the superior performance of ImageSI$_{text{DRTriplet}}$ and ImageSI$_{text{MDS}^{-1}}$ in terms of inference accuracy and interaction efficiency. Moreover, ImageSI$_{text{PHTriplet}}$ shows competitive results. The baseline model, WMDS$^{-1}$, generally exhibits lower performance metrics.

New fundamental observables are becoming accessible with the Leading Proton Spectrometers (LPS) of ZEUS and H1. This enables us to test more thoroughly the pQCD mechanism of diffractive Deep Inelastic Scattering (DIS). Calculations of the diffractive cross-section in the small Bjorken x limit have been performed. We have used the microscopic QCD formalism of diffractive DIS to find higher twist corrections to the transverse structure functions and predict the diffractive slope and azimuthal asymmetries. We establish duality correspondence between diffraction into low-mass continuum and vector meson production, and calculate the diffractive contribution to the spin structure functions. / Ph. D.

A major limitation in tissue engineering is the lack of nondestructive methods to assess the development of tissue scaffolds undergoing preconditioning in bioreactors. Due to significant optical scattering in most scaffolding materials, current microscope-based imaging methods cannot "see" through thick and optically opaque tissue constructs. To address this deficiency, we developed a scanning fiber imaging method capable of nondestructive imaging of fluorescently labeled cells through a thick and optically opaque vascular scaffold, contained in a bioreactor. This imaging modality is based on local excitation of fluorescent cells, acquisition of fluorescence through the scaffold, and fluorescence mapping based on the position of the excitation light. To evaluate the capability and accuracy of the imaging system, human endothelial cells, stably expressing green fluorescent protein (GFP), were imaged through a fibrous scaffold. Without sacrificing the scaffolds, we nondestructively visualized the distribution of GFP-labeled endothelial cells on the luminal surface through a ~500 µm thick tubular scaffold at cell-level resolutions and distinct localization. These results were similar to control images obtained using an optical microscope with direct line-of-sight access. Through a detailed quantitative analysis, we demonstrated that this method achieved a resolution of the order of 20-30 µm, with 10% or less deviation from standard optical microscopy. Furthermore, we demonstrated that the penetration depth of the imaging method exceeded that of confocal laser scanning microscopy by more than a factor of 2. Our imaging method also possesses a working distance (up to 8 cm) much longer than that of a standard confocal microscopy system, which can significantly facilitate bioreactor integration. This method will enable nondestructive monitoring of endothelial cells seeded on the lumen of a tissue-engineered vascular graft during preconditioning in vitro, as well as for other tissue-engineered constructs in the future. / Ph. D.

The abilities of powerful regression and multi-type data processing allow deep learning to effectively and accurately complete multi-tasks, which is the need of civil engineering. More cases showed that deep learning has become a greatly powerful and increasingly popular tool for civil engineering. Based on these, this dissertation developed deep learning studies for the condition assessment of civil infrastructure and materials. This dissertation included five main works: (1) Deep learning and infrared thermography for asphalt pavement crack severity classification. This work focused on longitudinal or transverse cracking. This work first built a dataset with four severity levels (no, low-severity, medium-severity, and high-severity) and three image types (visible, infrared, and fusion). Then this work applied the convolutional neural network (CNN) to classify the crack severity based on two strategies deep learning from scratch and transfer learning). This work also investigated the effect of image types on the accuracy of these two strategies and on the classification of different severity levels. (2) Asphalt pavement crack detection based on convolutional neural network and infrared thermography. This work first built an open dataset with three image types (visible, infrared, and fusion) and different conditions (single, multi, thin, and thick cracks; clean, rough, light, and dark backgrounds) and periods (morning, noon, and dusk). Then this work evaluated the performance of the CNN model based on the accuracy and complexity (computational and model). (3) An artificial neural network model on tensile behavior of hybrid steel-PVA fiber reinforced concrete containing fly ash and slag powder. This work considered a total of 23 factors for predicting the tensile behavior of hybrid fiber reinforced concrete (HFRC), including fibers' characteristics, mechanical properties of plain concrete, and concrete composition. Then this work compared the performance of the artificial neural network (ANN) method and the traditional equation-based method in terms of predicting the tensile stress, tensile strength, and strain corresponding to tensile strength. (4) Deep transfer learning-based vehicle classification by asphalt pavement vibration. This work first applied the pavement vibration IoT monitoring system to collect raw vibration signals and performed the wavelet transform to obtain denoised vibration signals. Then this work represented the vibration signals in two different ways, including the time-domain graph and the time-frequency graph. Finally, this work proposed two deep transfer learning-based vehicle classification methods according to these two representations of vibration signals. (5) Physical-informed long short-term memory (PI-LSTM) network for data-driven structural response modeling. This work first applied the single-degree-of-freedom (SDOF) system to investigate the performance of the proposed PI-LSTM network compared with the existing methods. Then this work further investigated and validated the proposed PI-LSTM network in terms of the experimental results of one six-story building and the numerical simulation results of another six-story building. / Doctor of Philosophy / With the development of technologies, deep learning has been applied to numerous fields to improve accuracy and efficiency. More work shows that deep learning has become a greatly powerful and increasingly popular tool for civil engineering. Since civil infrastructure and materials play a dominant role in civil engineering, this dissertation applied deep learning to the condition assessment of civil infrastructure and materials. Deep learning methods were applied to detect cracks in asphalt pavements. The mechanical properties of fiber reinforced concrete were investigated by deep learning methods. Based on the asphalt pavement vibration, the type of vehicles was classified by deep learning methods. Deep learning methods were also used to investigate the structural response.

In this work, we present different deep learning-based digital pre-distorters and compare them based on their performance towards improving the linearity of highly non-linear power amplifiers. The simulation results show that BiLSTM based DPDs work the best in terms of improving the linearity performance. We also compare two methodologies of direct learning and indirect learning to develop deep learning-based digital pre-distorters (DL-DPDs) models and evaluate their improvement on the linearity of Power Amplifiers (PA). We carry out a theoretical analysis on the differences between these training methodologies and verify their performance with simulation results on class-AB and class-F⁻¹ PAs. The simulation results show that both the learning methods lead to an improvement of more than 12 dB and 11dB in the linearity of class-AB and class-F⁻¹ PAs respectively, with indirect learning DL-DPD offering marginally better performance. Moreover, we compare the DL-DPD with memory polynomial models and show that using the former gives a significant improvement over the memory polynomials. Furthermore, we discuss the advantages of exploiting a BiLSTM based neural network architecture for designing direct/indirect DPDs. We demonstrate that BiLSTM DPD can be used to pre distort signals of any size without the drop in linearity. Moreover, based on the insights we develop a frequency domain loss using which further increased the linearity of the PA. / Master of Science / Wireless communication devices have fundamentally changed the way we interact with people. This increased the user's reliance on communication devices and significantly grew the need for higher data rates and faster internet speeds. But one major obstacle inside the transmitter chain (antenna) with increasing the data rates is the power amplifier, which distorts the signals at these higher powers. This distortion will reduce the efficiency and reliability of communication systems, greatly decreasing the quality of communication. So, we developed a high-performance DPD using deep learning to combat this issue. In this paper, we compare different deep learning-based DPDs and analyze which offers better performance. We also contrast two training methodologies to learn these DL-DPDs, theoretically and with simulation to arrive at which method offers better performing DPDs. We do these experiments on two different types of power amplifiers, and signals of any length. We design a new loss function, such that optimizing it leads to better DL-DPDs.

Medical images are extremely complicated to comprehend for a person without expertise. The limited number of practitioners across the globe often face the issue of fatigue due to the high number of cases. This fatigue, physical and mental, can induce human-errors during the diagnosis. In such scenarios, having an additional opinion can be helpful in boosting the confidence of the decision-maker. Thus, it becomes crucial to have a reliable Visual Question Answering (VQA) system which can provide a "second opinion" on medical cases. However, most of the VQA systems that work today cater to real-world problems and are not specifically tailored for handling medical images. Moreover, the VQA system for medical images needs to consider a limited amount of training data available in this domain. In this thesis, we develop a deep learning-based model for VQA on medical images taking the associated challenges into account. Our MedFuseNet system aims at maximizing the learning with minimal complexity by breaking the problem statement into simpler tasks and weaving everything together to predict the answer. We tackle two types of answer prediction - categorization and generation. We conduct an extensive set of both quantitative and qualitative analyses to evaluate the performance of MedFuseNet. Our results conclude that MedFuseNet outperforms other state-of-the-art methods available in the literature for these tasks. / Master of Science / Medical images are extremely complicated to comprehend for a person without expertise. The limited number of practitioners across the globe often face the issue of fatigue due to the high number of cases. This fatigue, physical and mental, can induce human-errors during the diagnosis. In such scenarios, having an additional opinion can be helpful in boosting the confidence of the decision-maker. Thus, it becomes crucial to have a reliable Visual Question Answering (VQA) system which can provide a "second opinion" on medical cases. However, most of the VQA systems that work today cater to real-world problems and are not specifically tailored for handling medical images. In this thesis, we propose an end-to-end deep learning-based system, MedFuseNet, for predicting the answer for the input query associated with the image. We cater to close-ended as well as open-ended type question-answer pairs. We conduct an extensive analysis to evaluate the performance of MedFuseNet. Our results conclude that MedFuseNet outperforms other state-of-the-art methods available in the literature for these tasks.

In the thesis, we aimed at building a robust UAV(drone) detection algorithm through which, one drone could detect another drone in flight. Though this was a straight forward object detection problem, the biggest challenge we faced for drone detection is the limited amount of drone images for training. To address this issue, we used Generative Adversarial Networks, CycleGAN to be precise, for the generation of realistic looking fake images which were indistinguishable from real data. CycleGAN is a classic example of Image to Image Translation technique, and we this applied in our situation where synthetic images from one domain were transformed into another domain, containing real data. The model, once trained, was capable of generating realistic looking images from synthetic data without the presence of real images. Following this, we employed a state of the art object detection model, YOLO(You Only Look Once), to build a Drone Detection model that was trained on the generated images. Finally, the performance of this model was compared against different datasets in order to evaluate its performance. / Master of Science / In the recent years, technologies like Deep Learning and Machine Learning have seen many rapid developments. Among the many applications they have, object detection is one of the widely used application and well established problems. In our thesis, we deal with a scenario where we have a swarm of drones and our aim is for one drone to recognize another drone in its field of vision. As there was no drone image dataset readily available, we explored different ways of generating realistic data to address this issue. Finally, we proposed a solution to generate realistic images using Deep Learning techniques and trained an object detection model on it where we evaluated how well it has performed against other models.

Organic soils are formed as deceased plant and animal wildlife is deposited and decomposed in wet environs. These soils have loose structures, low undrained strengths, and high natural water contents, and require improvement before they can be used as foundation materials. Previous researchers have found that the deep mixing method effectively improves organic soils. This study presents a quantitative and reliable method for predicting the strength of one organic soil treated with deep mixing. For this thesis, organic soils were manufactured from commercially available components. Soil-binder mixture specimens with different values of organic matter content, OM, binder content, water-to-binder ratio, and curing time were tested for unconfined compressive strength (UCS). Least-squares regression was used to fit a predictive equation, modified from the findings of previous researchers, to this data. The equation estimates the UCS of a deep-mixed organic soil specimen using its total water-to-binder ratio and mixture dry unit weight. Soil OM is incorporated into the equation as a threshold binder content, aT, required to improve a soil with a given OM; the aT term is used to calculate an effective total water-to-binder ratio. This thesis reached several important conclusions. The modified equation was successfully fitted to the data, meaning that the UCS of some organic soil-binder mixtures may be predicted in the same manner as that of inorganic soil-binder mixtures. The fitting coefficients from the predictive equations indicated that for the soils and binder tested, specimens of organic soil-binder mixtures have a greater relative gain of UCS immediately after mixing compared to specimens of inorganic soil-binder mixtures. However, the inorganic mixtures generally have a greater relative gain of UCS during the curing period. The influence of curing temperature was found to be similar for organic and inorganic mixtures. For the organic soils and binder tested in this research, aT may be expressed as a linear or power function of OM. For both functions, the value of aT was negligible at values of OM below 45%, which reflects the chemistry of the organic matter in the peat moss. For projects involving deep mixing of organic soils, the predictive equation will be used most effectively by fitting it to the results of bench-scale testing and then checking it against the results of field-scale testing. / Master of Science / Organic soils are formed continuously as matter from deceased organisms – mainly plants – is deposited in wet environs and decomposes. Organic soils are most commonly found in swamps, marshes, and coastal areas. These soils make poor foundation materials due to their low strengths. Deep mixing, or soil mixing, involves introducing a binder like Portland cement or lime into soil and blending the soil and binder together to form columns or blocks. Upon mixing, cementitious reactions occur, and the soil-binder mixture gains strength as it cures. Deep mixing may be performed using either a dry binder, known as dry mixing, or a binder-water slurry, referred to as wet mixing. Deep mixing may be used to treat either inorganic or organic soils to depths of 30 meters or greater. Contractor experience has shown that deep mixing is one of the most effective methods of improving the strength of organic soils. Lab-scale studies (by previous researchers) of wet mixing of inorganic soils have found that the strength of soil-binder mixtures can be expressed as a function of mixture curing time and curing temperature, as well as the quantity of binder used, or binder factor, and the consistency of the binder slurry. No corresponding expression has been generated for wet mixing of organic soils, although many studies on the subject have been performed by previous researchers. The goal of this research was to generate such an expression for one organic soil. The soil used was made of sphagnum peat moss, an organic material commonly found in nature, and an inorganic clay used by previous researchers in studies of deep mixing in inorganic soils. The binder used in this research was a Portland cement. For this research, 43 unique soil-binder mixtures were manufactured. Each mixture involved a unique combination of soil organic matter content, binder factor, and binder slurry consistency. After a soil-binder mixture was made, it was divided, placed into cylindrical molds, and allowed to cure. The temperature of the curing environment of the mixture was monitored. Mixture compressive strength was assessed after 7, 14, and 28 days of curing using two cylindrically molded specimens of the mixture. Data on mixture strength was then evaluated to assess whether it could be expressed as a function of the variables tested. iv This research determined that the strength of at least some organic soils improved with wet mixing can be expressed as a function of soil organic matter content, binder factor, binder slurry consistency, and mixture curing time and curing temperature. The function will likely prove useful to deep mixing contractors, who routinely perform lab-scale deep mixing trials on samples of the soils to be improved in the field. Assuming wet mixing is used, the results of the trials are used to select values of binder factor and binder slurry consistency for the project. The function generated from this research will allow deep mixing contractors to select these values more reliably during the lab-scale phase of their work.