Depth Estimation in Stereo Biomedical Images via Proxy-Supervised Deep Learning

In order to estimate depth through supervised deep learning-based stereo methods, it is necessary to have access to precise ground truth depth data. While the gathering of precise labels is commonly tackled by deploying depth sensors, this is not always a viable solution. For instance, in many applications in the biomedical domain, the choice of sensors capable of sensing depth at small distances with high precision on difficult surfaces (that present non-Lambertian properties) is very limited. It is therefore necessary to find alternative techniques to gather ground truth data without having to rely on external sensors. In this thesis, two different approaches have been tested to produce supervision data for biomedical images. The first aims to obtain input stereo image pairs and disparities through simulation in a virtual environment, while the second relies on a non-learned disparity estimation algorithm in order to produce noisy disparities, which are then filtered by means of hand-crafted confidence measures to create noisy labels for a subset of pixels. Among the two, the second approach, which is referred in literature as proxy-labeling, has shown the best results and has even outperformed the non-learned disparity estimation algorithm used for supervision.

Identificazione di caratteristiche rilevanti in capi di abbigliamento mediante deep-learning

Il mondo della moda è in continua e costante evoluzione, non solo dal punto di vista sociale, ma anche da quello tecnologico. Nel corso del presente elaborato si è studiata la possibilità di riconoscere e segmentare abiti presenti in una immagine utilizzando reti neurali profonde e approcci moderni. Sono state, quindi, analizzate reti quali FasterRCNN, MaskRCNN, YOLOv5, FashionPedia e Match-RCNN. In seguito si è approfondito l’addestramento delle reti neurali profonde in scenari di alta parallelizzazione e su macchine dotate di molteplici GPU al fine di ridurre i tempi di addestramento. Inoltre si è sperimentata la possibilità di creare una rete per prevedere se un determinato abito possa avere successo in futuro analizzando semplicemente dati passati e una immagine del vestito in questione. Necessaria per tali compiti è stata, inoltre, una approfondita analisi dei dataset esistenti nel mondo della moda e dei metodi per utilizzarli per l’addestramento. Il presente elaborato è stato svolto nell’ambito del progetto FA.RE.TRA. per il quale l'Università di Bologna svolge un compito di consulenza per lo studio di fattibilità su reti neurali in grado di svolgere i compiti menzionati.

Convolutional Neural Network Architectures for Template Matching

The Neural Networks customized and tested in this thesis (WaldoNet, FlowNet and PatchNet) are a first exploration and approach to the Template Matching task. The possibilities of extension are therefore many and some are proposed below. During my thesis, I have analyzed the functioning of the classical algorithms and adapted with deep learning algorithms. The features extracted from both the template and the query images resemble the keypoints of the SIFT algorithm. Then, instead of similarity function or keypoints matching, WaldoNet and PatchNet use the convolutional layer to compare the features, while FlowNet uses the correlational layer. In addition, I have identified the major challenges of the Template Matching task (affine/non-affine transformations, intensity changes...) and solved them with a careful design of the dataset.

Mitigating non-Lambertian surfaces issues in Stereo Matching with Neural Radiance Fields

Depth estimation from images has long been regarded as a preferable alternative compared to expensive and intrusive active sensors, such as LiDAR and ToF. The topic has attracted the attention of an increasingly wide audience thanks to the great amount of application domains, such as autonomous driving, robotic navigation and 3D reconstruction. Among the various techniques employed for depth estimation, stereo matching is one of the most widespread, owing to its robustness, speed and simplicity in setup. Recent developments has been aided by the abundance of annotated stereo images, which granted to deep learning the opportunity to thrive in a research area where deep networks can reach state-of-the-art sub-pixel precision in most cases. Despite the recent findings, stereo matching still begets many open challenges, two among them being finding pixel correspondences in presence of objects that exhibits a non-Lambertian behaviour and processing high-resolution images. Recently, a novel dataset named Booster, which contains high-resolution stereo pairs featuring a large collection of labeled non-Lambertian objects, has been released. The work shown that training state-of-the-art deep neural network on such data improves the generalization capabilities of these networks also in presence of non-Lambertian surfaces. Regardless being a further step to tackle the aforementioned challenge, Booster includes a rather small number of annotated images, and thus cannot satisfy the intensive training requirements of deep learning. This thesis work aims to investigate novel view synthesis techniques to augment the Booster dataset, with ultimate goal of improving stereo matching reliability in presence of high-resolution images that displays non-Lambertian surfaces.

Machine learning and convolutional neural networks for melanomas identification

Il machine learning negli ultimi anni ha acquisito una crescente popolarità nell’ambito della ricerca scientifica e delle sue applicazioni. Lo scopo di questa tesi è stato quello di studiare il machine learning nei suoi aspetti generali e applicarlo a problemi di computer vision. La tesi ha affrontato le difficoltà del dover spiegare dal punto di vista teorico gli algoritmi alla base delle reti neurali convoluzionali e ha successivamente trattato due problemi concreti di riconoscimento immagini: il dataset MNIST (immagini di cifre scritte a mano) e un dataset che sarà chiamato ”MELANOMA dataset” (immagini di melanomi e nevi sani). Utilizzando le tecniche spiegate nella sezione teorica si sono riusciti ad ottenere risultati soddifacenti per entrambi i dataset ottenendo una precisione del 98% per il MNIST e del 76.8% per il MELANOMA dataset

Neural Radiance Fields for Relighting and View Synthesis

Neural scene representation and neural rendering are new computer vision techniques that enable the reconstruction and implicit representation of real 3D scenes from a set of 2D captured images, by fitting a deep neural network. The trained network can then be used to render novel views of the scene. A recent work in this field, Neural Radiance Fields (NeRF), presented a state-of-the-art approach, which uses a simple Multilayer Perceptron (MLP) to generate photo-realistic RGB images of a scene from arbitrary viewpoints. However, NeRF does not model any light interaction with the fitted scene; therefore, despite producing compelling results for the view synthesis task, it does not provide a solution for relighting. In this work, we propose a new architecture to enable relighting capabilities in NeRF-based representations and we introduce a new real-world dataset to train and evaluate such a model. Our method demonstrates the ability to perform realistic rendering of novel views under arbitrary lighting conditions.

Generative Neural Networks for image iper-resolution and improvement of Optical Character Recognition's performances

The usage of Optical Character Recognition’s (OCR, systems is a widely spread technology into the world of Computer Vision and Machine Learning. It is a topic that interest many field, for example the automotive, where becomes a specialized task known as License Plate Recognition, useful for many application from the automation of toll road to intelligent payments. However, OCR systems need to be very accurate and generalizable in order to be able to extract the text of license plates under high variable conditions, from the type of camera used for acquisition to light changes. Such variables compromise the quality of digitalized real scenes causing the presence of noise and degradation of various type, which can be minimized with the application of modern approaches for image iper resolution and noise reduction. Oneclass of them is known as Generative Neural Networks, which are very strong ally for the solution of this popular problem.

Antibody-Antigen Interface Characterization with Contextualized Pre-trained Embeddings

Artificial Intelligence (AI) has substantially influenced numerous disciplines in recent years. Biology, chemistry, and bioinformatics are among them, with significant advances in protein structure prediction, paratope prediction, protein-protein interactions (PPIs), and antibody-antigen interactions. Understanding PPIs is critical since they are responsible for practically everything living and have several uses in vaccines, cancer, immunology, and inflammatory illnesses. Machine Learning (ML) offers enormous potential for effectively simulating antibody-antigen interactions and improving in-silico optimization of therapeutic antibodies for desired features, including binding activity, stability, and low immunogenicity. This research looks at the use of AI algorithms to better understand antibody-antigen interactions, and it further expands and explains several difficulties encountered in the field. Furthermore, we contribute by presenting a method that outperforms existing state-of-the-art strategies in paratope prediction from sequence data.

Continual learning on nano drones

Miniaturized flying robotic platforms, called nano-drones, have the potential to revolutionize the autonomous robots industry sector thanks to their very small form factor. The nano-drones’ limited payload only allows for a sub-100mW microcontroller unit for the on-board computations. Therefore, traditional computer vision and control algorithms are too computationally expensive to be executed on board these palm-sized robots, and we are forced to rely on artificial intelligence to trade off accuracy in favor of lightweight pipelines for autonomous tasks. However, relying on deep learning exposes us to the problem of generalization since the deployment scenario of a convolutional neural network (CNN) is often composed by different visual cues and different features from those learned during training, leading to poor inference performances. Our objective is to develop and deploy and adaptation algorithm, based on the concept of latent replays, that would allow us to fine-tune a CNN to work in new and diverse deployment scenarios. To do so we start from an existing model for visual human pose estimation, called PULPFrontnet, which is used to identify the pose of a human subject in space through its 4 output variables, and we present the design of our novel adaptation algorithm, which features automatic data gathering and labeling and on-device deployment. We therefore showcase the ability of our algorithm to adapt PULP-Frontnet to new deployment scenarios, improving the R2 scores of the four network outputs, with respect to an unknown environment, from approximately [−0.2, 0.4, 0.0,−0.7] to [0.25, 0.45, 0.2, 0.1]. Finally we demonstrate how it is possible to fine-tune our neural network in real time (i.e., under 76 seconds), using the target parallel ultra-low power GAP 8 System-on-Chip on board the nano-drone, and we show how all adaptation operations can take place using less than 2mWh of energy, a small fraction of the available battery power.

Framework for anomaly detection on photovoltaic plants

Technological advancement has undergone exponential growth in recent years, and this has brought significant improvements in the computational capabilities of computers, which can now perform an enormous amount of calculations per second. Taking advantage of these improvements has made it possible to devise algorithms that are very demanding in terms of the computational resources needed to develop architectures capable of solving the most complex problems: currently the most powerful of these are neural networks and in this thesis I will combine these tecniques with classical computer vision algorithms to improve the speed and accuracy of maintenance in photovoltaic facilities.

Suicide in urban New South Wales, Australia 1985-1994: Socio-economic and migrant interactions

Variation of suicide with socio-economic status (SES) in urban NSW (Australia) during 1985-1994, by sex and country or region of birth, was examined using Poisson regression analysis of vital statistics and population data (age greater than or similar to 15 yr). Quintiles of SES were defined by municipality of residence and comparisons of suicide by SES were adjusted for age and country (or region) of birth (COB), and examined by COB. Risk of suicide in females was 28% that of males for all adults and 21% for youth (age 15-24 yr). Suicide risk was lower in males from southern Europe, Middle East and Asia, and higher in northern and eastern European males, compared to the Australian-born. Risks for suicide increased significantly with decreasing SES in males, but not in females. The relationship of male suicide and SES was stronger when controlled for COB. For males, the relative risk of suicide, adjusted for age and COB, was 66% higher in the lowest SES quintile compared to the highest quintile, and 39% higher for youth (age 15-24 yr). For male suicide, the population attributable fraction for SES (less than the highest quintile) was 27%. Analysis of SES differentials in male suicide according to COB indicated a significant inverse suicide gradient in relation to SES for the Australian-born and those burn in New Zealand and the United Kingdom or fire. but not in non-English speaking COB groups, except for Asia. For Australian-born males, suicide risk was 71% higher in the lowest SES group (compared to the highest), adjusted for age. These findings indicate that SES plays an important role in male suicide rates among the Australian-born and migrants from English-speaking countries and Asia, and among youth; but not in female suicide, nor suicide in most non-English speaking migrant groups. Reduction in SES differentials through economic and social policies may reduce male suicide in lower SES groups and should be seen to be at least as important as individual level interventions. (C) 1998 Elsevier Science Ltd. All rights reserved.

Utilisation de la coronarographie en Suisse: une étude de population. [The use of coronary vessel radiography in Switzerland: a population study].

All patients having undergone a coronarography during 1984 have been surveyed in Switzerland. This retrospective study has used existing data in the 13 centers practicing this diagnostic procedure. 4921 coronarographies were carried out in 1984, amongst 4359 patients. In terms of population-based rates, the national figures are 77 procedures/100,000 residents, and 68 patients/100,000 residents. Female rates are one fourth of the male rates (27/100,000 versus 112/100,000). For both sexes, the highest utilization rates are for the age groups 60-64. Swiss figures are relatively low when compared with other developed countries. However, patterns of utilization are very different within the country: according to the Canton of residence of the patient, the utilization rates (standardized for age and sex) vary from 8/100,000 to 160/100,000. There is a distinct gradient from south-west to north-east, which closely corresponds to the distribution of centers practicing the procedure. More intriguing is the fact that cardiovascular mortality shows an inverse geographical gradient, with the highest mortality in Cantons having the lowest rate of coronarography. Various reasons for the observed variations are discussed, in relation with differences in supply of diagnostic and therapeutic equipments, but also in relation with various patterns of demand related to differential morbidity rates and/or differential patterns of clinical decision.

RF-sputter fabrication of magnetic garnet thin films and simulation modeling for 1-D magnetic photonic crystal waveguide devices

One dimensional magnetic photonic crystals (1D-MPC) are promising structures for integrated optical isolator applications. Rare earth substituted garnet thin films with proper Faraday rotation are required to fabricate planar 1D-MPCs. In this thesis, flat-top response 1D-MPC was proposed and spectral responses and Faraday rotation were modeled. Bismuth substituted iron garnet films were fabricated by RF magnetron sputtering and structures, compositions, birefringence and magnetooptical properties were studied. Double layer structures for single mode propagation were also fabricated by sputtering for the first time. Multilayer stacks with multiple defects (phase shift) composed of Ce-YIG and GGG quarter-wave plates were simulated by the transfer matrix method. The transmission and Faraday rotation characteristics were theoretically studied. It is found that flat-top response, with 100% transmission and near 45o rotation is achievable by adjusting the inter-defect spacing, for film structures as thin as 30 to 35 μm. This is better than 3-fold reduction in length compared to the best Ce-YIG films for comparable rotations, thus allows a considerable reduction in size in manufactured optical isolators. Transmission bands as wide as 7nm were predicted, which is considerable improvement over 2 defects structure. Effect of repetition number and ratio factor on transmission and Faraday rotation ripple factors for the case of 3 and 4 defects structure has been discussed. Diffraction across the structure corresponds to a longer optical path length. Thus the use of guided optics is required to minimize the insertion losses in integrated devices. This part is discussed in chapter 2 in this thesis. Bismuth substituted iron garnet thin films were prepared by RF magnetron sputtering. We investigated or measured the deposition parameters optimization, crystallinity, surface morphologies, composition, magnetic and magnetooptical properties. A very high crystalline quality garnet film with smooth surface has been heteroepitaxially grown on (111) GGG substrate for films less than 1μm. Dual layer structures with two distinct XRD peaks (within a single sputtered film) start to develop when films exceed this thickness. The development of dual layer structure was explained by compositional gradient across film thickness, rather than strain gradient proposed by other authors. Lower DC self bias or higher substrate temperature is found to help to delay the appearance of the 2nd layer. The deposited films show in-plane magnetization, which is advantageous for waveguide devices application. Propagation losses of fabricated waveguides can be decreased by annealing in an oxygen atmosphere from 25dB/cm to 10dB/cm. The Faraday rotation at λ=1.55μm were also measured for the waveguides. FR is small (10° for a 3mm long waveguide), due to the presence of linear birefringence. This part is covered in chapter 4. We also investigated the elimination of linear birefringence by thickness tuning method for our sputtered films. We examined the compressively and tensilely strained films and analyze the photoelastic response of the sputter deposited garnet films. It has been found that the net birefringence can be eliminated under planar compressive strain conditions by sputtering. Bi-layer GGG on garnet thin film yields a reduced birefringence. Temperature control during the sputter deposition of GGG cover layer is critical and strongly influences the magnetization and birefringence level in the waveguide. High temperature deposition lowers the magnetization and increases the linear birefringence in the garnet films. Double layer single mode structures fabricated by sputtering were also studied. The double layer, which shows an in-plane magnetization, has an increased RMS roughness upon upper layer deposition. The single mode characteristic was confirmed by prism coupler measurement. This part is discussed in chapter 5.

Wind Shear, Gust, and Yaw-Induced Dynamic Stall on Wind-Turbine Blades

This study examined the effect of a spanwise angle of attack gradient on the growth and stability of a dynamic stall vortex in a rotating system. It was found that a spanwise angle of attack gradient induces a corresponding spanwise vorticity gradient, which, in combination with spanwise flow, results in a redistribution of circulation along the blade. Specifically, when modelling the angle of attack gradient experienced by a wind turbine at the 30% span position during a gust event, the spanwise vorticity gradient was aligned such that circulation was transported from areas of high circulation to areas of low circulation, increasing the local dynamic stall vortex growth rate, which corresponds to an increase in the lift coefficient, and a decrease in the local vortex stability at this point. Reversing the relative alignment of the spanwise vorticity gradient and spanwise flow results in circulation transport from areas of low circulation generation to areas of high circulation generation, acting to reduce local circulation and stabilise the vortex. This circulation redistribution behaviour describes a mechanism by which the fluctuating loads on a wind turbine are magnified, which is detrimental to turbine lifetime and performance. Therefore, an understanding of this phenomenon has the potential to facilitate optimised wind turbine design.

A gradient descent algorithm for minimizing amino acid coupling reactions when synthesizing cyclic-peptide libraries

Combinatorial chemistry has become an invaluable tool in medicinal chemistry for the identification of new drug leads. For example, libraries of predetermined sequences and head-to-tail cyclized peptides are routinely synthesized in our laboratory using the IRORI approach. Such libraries are used as molecular toolkits that enable the development of pharmacophores that define activity and specificity at receptor targets. These libraries can be quite large and difficult to handle, due to physical and chemical constraints imposed by their size. Therefore, smaller sub-libraries are often targeted for synthesis. The number of coupling reactions required can be greatly reduced if the peptides having common amino acids are grouped into the same sub-library (batching). This paper describes a schedule optimizer to minimize the number of coupling reactions by rotating and aligning sequences while simultaneously batching. The gradient descent method thereby reduces the number of coupling reactions required for synthesizing cyclic peptide libraries. We show that the algorithm results in a 75% reduction in the number of coupling reactions for a typical cyclic peptide library.