Image understanding is the task of interpreting images by effectively solving the individual tasks of object recognition and semantic image segmentation. An image understanding system must have the capacity to distinguish between similar looking image regions while being invariant in its response to regions that have been altered by the appearance-altering transformation. The fundamental challenge for any such system lies within this simultaneous requirement for both invariance and specificity. Many image understanding systems have been proposed that capture geometric properties such as shapes, textures, motion and 3D perspective projections using filtering, non-linear modulus, and pooling operations. Deep learning networks ignore these geometric considerations and compute descriptors having suitable invariance and stability to geometric transformations using (end-to-end) learned multi-layered network filters. These deep learning networks in recent years have come to dominate the previously separate fields of research in machine learning, computer vision, natural language understanding and speech recognition. Despite the success of these deep networks, there remains a fundamental lack of understanding in the design and optimization of these networks which makes it difficult to develop them. Also, training of these networks requires large labeled datasets which in numerous applications may not be available. In this dissertation, we propose the ScatterNet Hybrid Framework for Deep Learning that is inspired by the circuitry of the visual cortex. The framework uses a hand-crafted front-end, an unsupervised learning based middle-section, and a supervised back-end to rapidly learn hierarchical features from unlabelled data. Each layer in the proposed framework is automatically optimized to produce the desired computationally efficient architecture. The term `Hybrid' is coined because the framework uses both unsupervised as well as supervised learning. We propose two hand-crafted front-ends that can extract locally invariant features from the input signals. Next, two ScatterNet Hybrid Deep Learning (SHDL) networks (a generative and a deterministic) were introduced by combining the proposed front-ends with two unsupervised learning modules which learn hierarchical features. These hierarchical features were finally used by a supervised learning module to solve the task of either object recognition or semantic image segmentation. The proposed front-ends have also been shown to improve the performance and learning of current Deep Supervised Learning Networks (VGG, NIN, ResNet) with reduced computing overhead.

abstract: In recent years, conventional convolutional neural network (CNN) has achieved outstanding performance in image and speech processing applications. Unfortunately, the pooling operation in CNN ignores important spatial information which is an important attribute in many applications. The recently proposed capsule network retains spatial information and improves the capabilities of traditional CNN. It uses capsules to describe features in multiple dimensions and dynamic routing to increase the statistical stability of the network. In this work, we first use capsule network for overlapping digit recognition problem. We evaluate the performance of the network with respect to recognition accuracy, convergence and training time per epoch. We show that capsule network achieves higher accuracy when training set size is small. When training set size is larger, capsule network and conventional CNN have comparable recognition accuracy. The training time per epoch for capsule network is longer than conventional CNN because of the dynamic routing algorithm. An analysis of the GPU timing shows that adjusting the capsule structure can help decrease the time complexity of the dynamic routing algorithm significantly. Next, we design a capsule network for speech recognition, specifically, overlapping word recognition. We use both capsule network and conventional CNN to recognize 2 overlapping words in speech files created from 5 word classes. We show that capsule network achieves a considerably higher recognition accuracy (96.92%) compared to conventional CNN (85.19%). Our results show that capsule network recognizes overlapping word by recognizing each individual word in the speech. We also verify the scalability of capsule network by increasing the number of word classes from 5 to 10. Capsule network still shows a high recognition accuracy of 95.42% in case of 10 words while the accuracy of conventional CNN decreases sharply to 73.18%. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2018

Organismos de mar profundo encontram limitações na disponibilidade de alimentos e exploram enriquecimentos orgânicos esporádicos que chegam ao assoalho oceânico. O objetivo deste trabalho foi descrever a diversidade das comunidades bacterianas associadas a parcelas sintéticas e orgânicas (vértebras de baleia e blocos de madeira) no Espírito Santo, Rio de Janeiro e São Paulo a 3.300 m de profundidade, avaliando a influência dos substratos e da localização geográfica sobre essas comunidades. Foi realizada a extração de DNA e amplificação do gene RNAr 16S para sequenciamento por Illumina Miseq e análises estatísticas pelo Qiime. Os Gêneros dominantes nos substratos sintéticos, madeira e vértebras foram Psychroserpens (Flavobacteriia), Phaeobacter, (Alphaproteobacteria), Desulfobacter, (Deltaproteobacteria), respectivamente. Com base nos resultados obtidos, afirma-se que o tipo de substrato teve maior influência do que a localização geográfica sobre a estrutura das comunidades bacterianas. / Deep sea organisms found limitations in the availability of food and exploit sporadic organic enrichments that reach the ocean floor. The aim of this study was to describe the diversity of bacterial communities associated with synthetic and organic substrate (whale bone and wood blocks) in Espírito Santo, Rio de Janeiro and Sao Paulo to 3,300 m deep, assessing the influence of substrates and location geographical about these communities. 16S rRNA sequencing was performed by Illumina Miseq and statistical analysis by Qiime. The dominant genera in synthetic substrates, wood and vertebrae were Psychroserpens (Flavobacteriia), Phaeobacter (Alphaproteobacteria) and Desulfobacter, (Deltaproteobacteria), respectively. Based on these results, it is stated that the substrate type had greater influence than geographic location on the structure of bacterial communities.

Deep Learning is dominant in the field of computer vision, thanks to its high performance. This high performance is driven by large annotated datasets and proper evaluation benchmarks. However, two important areas in computer vision, depth-based hand segmentation, and local features, respectively lack a large well-annotated dataset and a benchmark protocol that properly demonstrates its practical performance. Therefore, in this thesis, we focus on these two problems. For hand segmentation, we create a novel systematic way to easily create automatic semantic segmentation annotations for large datasets. We achieved this with the help of traditional computer vision techniques and minimal hardware setup of one RGB-D camera and two distinctly colored skin-tight gloves. Our method allows easy creation of large-scale datasets with high annotation quality. For local features, we create a new modern benchmark, that reveals their different aspects. Specifically wide-baseline stereo matching and Multi-View Stereo (MVS), of keypoints in a more practical setup, namely Structure-from-Motion (SfM). We believe that through our new benchmark, we will be able to spur research on learned local features to a more practical direction. In this respect, the benchmark developed for the thesis will be used to host a challenge on local features. / Graduate

abstract: Non-line-of-sight (NLOS) imaging of objects not visible to either the camera or illumina- tion source is a challenging task with vital applications including surveillance and robotics. Recent NLOS reconstruction advances have been achieved using time-resolved measure- ments. Acquiring these time-resolved measurements requires expensive and specialized detectors and laser sources. In work proposes a data-driven approach for NLOS 3D local- ization requiring only a conventional camera and projector. The localisation is performed using a voxelisation and a regression problem. Accuracy of greater than 90% is achieved in localizing a NLOS object to a 5cm × 5cm × 5cm volume in real data. By adopting the regression approach an object of width 10cm to localised to approximately 1.5cm. To generalize to line-of-sight (LOS) scenes with non-planar surfaces, an adaptive lighting al- gorithm is adopted. This algorithm, based on radiosity, identifies and illuminates scene patches in the LOS which most contribute to the NLOS light paths, and can factor in sys- tem power constraints. Improvements ranging from 6%-15% in accuracy with a non-planar LOS wall using adaptive lighting is reported, demonstrating the advantage of combining the physics of light transport with active illumination for data-driven NLOS imaging. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2019

Computed tomography (CT) is routinely used for diagnosing lung disease and developing treatment plans using images of intricate lung structure with submillimeter resolution. Automated segmentation of anatomical structures in such images is important to enable efficient processing in clinical and research settings. Convolution neural networks (ConvNets) are largely successful at performing image segmentation with the ability to learn discriminative abstract features that yield generalizable predictions. However, constraints in hardware memory do not allow deep networks to be trained with high-resolution volumetric CT images. Restricted by memory constraints, current applications of ConvNets on volumetric medical images use a subset of the full image; limiting the capacity of the network to learn informative global patterns. Local patterns, such as edges, are necessary for precise boundary localization, however, they suffer from low specificity. Global information can disambiguate structures that are locally similar. The central thesis of this doctoral work is that both local and global information is important for segmentation of anatomical structures in medical images. A novel multi-scale ConvNet is proposed that divides the learning task across multiple networks; each network learns features over different ranges of scales. It is hypothesized that multi-scale ConvNets will lead to improved segmentation performance, as no compromise needs to be made between image resolution, image extent, and network depth. Three multi-scale models were designed to specifically target segmentation of three pulmonary structures: lungs, fissures, and lobes. The proposed models were evaluated on a diverse datasets and compared to architectures that do not use both local and global features. The lung model was evaluated on humans and three animal species; the results demonstrated the multi-scale model outperformed single scale models at different resolutions. The fissure model showed superior performance compared to both a traditional Hessian filter and a standard U-Net architecture that is limited in global extent. The results demonstrated that multi-scale ConvNets improved pulmonary CT segmentation by incorporating both local and global features using multiple ConvNets within a constrained-memory system. Overall, the proposed pipeline achieved high accuracy and was robust to variations resulting from different imaging protocols, reconstruction kernels, scanners, lung volumes, and pathological alterations; demonstrating its potential for enabling high-throughput image analysis in clinical and research settings.

A portion of teachers in the United States educational system don'€™t use data to inform and improve their instruction resulting in actionable change. A gap exists between teachers having and interpreting data and making meaning in such a way that leads to actionable change in instruction. The purpose of this case study was to investigate how teachers used data to alter instruction and identify factors that inhibited or supported teachers in using data to drive instructional practice. This study was guided by Ackoff'€™s theory of action cycle, which included interaction, dialogue, data discoveries, and team response to data. The research questions asked how teams used data and what factors inhibited and supported the use of data. Three teams were observed. Eleven classroom teachers, the building principal and the district professional development director were interviewed. The teacher team criteria included that teachers met weekly and used, at a minimum, common formative assessments. The school and district mission, vision and value statements were collected as artifacts to see how these documents supported the use of data. Open and axial coding exposed themes and patterns. Results indicated that teachers commonly omitted one or more phases in a data cycle; however, when teachers worked through all phases of a data cycle, actionable change in instruction resulted, and factors that both inhibited and supported teacher use of data to guide instruction were evident throughout all aspects of the study. The project, a white paper, summarized the study and provided research-based recommendations based on the study. These recommendations focus on building teacher capacity and relationships. This study may generate social change through educational equity. Equity is achieved when teachers use data to inform instruction so that learners of all abilities may have access to learning.

This study investigates Mio-Pliocene mass-transport deposits (MTDs) in an understudied, hydrocarbon-rich region of the northeastern Gulf of Mexico. The research utilizes a high-quality 3D seismic dataset with an area of 635 km2, along with wireline logs and biostratigraphic data. With the help of quantitative seismic geomorphology techniques, detailed mapping of MTDs suggests a complex erosional and depositional history. Deposition of a MTD unit resulted in a 180 m topographic high that substantially influenced the distribution and morphology of subsequent MTDs, specifically the bifurcation of later mass-transport flows. This bifurcation contributed to the generation of a non-shielded erosional remnant with an area of 65 km2. Depositional elements of the remnant strata are interpreted to be sediment waves. Instantaneous frequency attribute maps of the erosional remnant suggest a different lithology than the surrounding muddy MTDs; and, thus, the remnant unit is interpreted to be sandy. For the first time in literature, this research documented intra-MTD channel and lobe features. The development of a sinuous channel system encased within MTD gives new insights into mass-transport processes. This provides evidence for considering MTD as amalgamation deposits of multiple and different-type of flow events (e.g., turbidity currents and debris flows), rather than a singular event-deposit. The channel, lobe, and erosional remnant features examined in this research demonstrate reservoir-prone facies encased within MTD units, forming stratigraphic traps directly associated with mass-transport phenomena. This research contributes to the understanding of seal vs. reservoir rock development and distribution in the study area, as well as presents new developments into mass-transport deposit flow processes and their resulting morphologies.

The ultimate goal of image understanding is to transfer visual images into numerical or symbolic descriptions of the scene that are helpful for decision making. Knowing when, where, and in which direction a picture was taken, the task of geo-calibration makes it possible to use imagery to understand the world and how it changes in time. Current models for geo-calibration are mostly deterministic, which in many cases fails to model the inherent uncertainties when the image content is ambiguous. Furthermore, without a proper modeling of the uncertainty, subsequent processing can yield overly confident predictions. To address these limitations, we propose a probabilistic model for camera geo-calibration using deep neural networks. While our primary contribution is geo-calibration, we also show that learning to geo-calibrate a camera allows us to implicitly learn to understand the content of the scene.

<div>Following a natural disaster, one of the most important facet that influence a persons chances of survival/being found out is the time with which they are rescued. Traditional means of search operations involving dogs, ground robots, humanitarian intervention; are time intensive and can be a major bottleneck in search operations. The main aim of these operations is to rescue victims without critical delay in the shortest time possible which can be realized in real-time by using UAVs. With advancements in computational devices and the ability to learn from complex data, deep learning can be leveraged in real time environment for purpose of search and rescue operations. This research aims to solve the traditional means of search operation using the concept of deep learning for real time object detection and Photogrammetry for precise geo-location mapping of the objects(person,car) in real time. In order to do so, various pre-trained algorithms like Mask-RCNN, SSD300, YOLOv3 and trained algorithms like YOLOv3 have been deployed with their results compared with means of addressing the search operation in</div><div>real time.</div><div><br></div>