Implementazione di banca dati di immagini per lo sviluppo di algoritmi per la navigazione aerea automatica

Lo scopo di questo studio è l’implementazione di un sistema di navigazione autonomo in grado di calcolare la traiettoria di un mezzo aereo, noti che siano a priori dei punti di posizione detti waypoint. A partire da questa traiettoria, è possibile ottenere la sua rappresentazione in un codice che mette a disposizione immagini satellitari e ricavare le viste del terreno sorvolato in una serie di punti calcolati, in modo da garantire in ogni sequenza la presenza di elementi comuni rispetto a quella precedente. Lo scopo della realizzazione di questa banca dati è rendere possibili futuri sviluppi di algoritmi di navigazione basati su deep learning e reti neurali. Le immagini virtuali ottenute del terreno saranno in futuro applicate alla navigazione autonoma per agricoltura di precisione mediante droni. Per lo studio condotto è stato simulato un generico velivolo, con o senza pilota, dotato di una videocamera fissata su una sospensione cardanica a tre assi (gimbal). La tesi, dunque, introduce ai più comuni metodi di determinazione della posizione dei velivoli e alle più recenti soluzioni basate su algoritmi di Deep Learning e sistemi vision-based con reti neurali e segue in un approfondimento sul metodo di conversione degli angoli e sulla teoria matematica che ne sta alla base. Successivamente, analizza nel dettaglio il processo di simulazione della navigazione autonoma e della determinazione della traiettoria in ambiente software Matlab e Simulink, procedendo nell’analisi di alcuni casi di studio in ambienti realistici. L’elaborato si conclude con un breve riepilogo di quanto svolto e con alcune considerazioni sugli sviluppi futuri.

Computer vision in aseptic pharmaceutical manufacturing: a deep learning-based approach to vial inspection and pack formation during a lyophilizer loading cycle

Vision systems are powerful tools playing an increasingly important role in modern industry, to detect errors and maintain product standards. With the enlarged availability of affordable industrial cameras, computer vision algorithms have been increasingly applied in industrial manufacturing processes monitoring. Until a few years ago, industrial computer vision applications relied only on ad-hoc algorithms designed for the specific object and acquisition setup being monitored, with a strong focus on co-designing the acquisition and processing pipeline. Deep learning has overcome these limits providing greater flexibility and faster re-configuration. In this work, the process to be inspected consists in vials’ pack formation entering a freeze-dryer, which is a common scenario in pharmaceutical active ingredient packaging lines. To ensure that the machine produces proper packs, a vision system is installed at the entrance of the freeze-dryer to detect eventual anomalies with execution times compatible with the production specifications. Other constraints come from sterility and safety standards required in pharmaceutical manufacturing. This work presents an overview about the production line, with particular focus on the vision system designed, and about all trials conducted to obtain the final performance. Transfer learning, alleviating the requirement for a large number of training data, combined with data augmentation methods, consisting in the generation of synthetic images, were used to effectively increase the performances while reducing the cost of data acquisition and annotation. The proposed vision algorithm is composed by two main subtasks, designed respectively to vials counting and discrepancy detection. The first one was trained on more than 23k vials (about 300 images) and tested on 5k more (about 75 images), whereas 60 training images and 52 testing images were used for the second one.

Graph-based Vision-Language Model for Multi-Modal Retrieval of Fashion Products

Artificial Intelligence is reshaping the field of fashion industry in different ways. E-commerce retailers exploit their data through AI to enhance their search engines, make outfit suggestions and forecast the success of a specific fashion product. However, it is a challenging endeavour as the data they possess is huge, complex and multi-modal. The most common way to search for fashion products online is by matching keywords with phrases in the product's description which are often cluttered, inadequate and differ across collections and sellers. A customer may also browse an online store's taxonomy, although this is time-consuming and doesn't guarantee relevant items. With the advent of Deep Learning architectures, particularly Vision-Language models, ad-hoc solutions have been proposed to model both the product image and description to solve this problems. However, the suggested solutions do not exploit effectively the semantic or syntactic information of these modalities, and the unique qualities and relations of clothing items. In this work of thesis, a novel approach is proposed to address this issues, which aims to model and process images and text descriptions as graphs in order to exploit the relations inside and between each modality and employs specific techniques to extract syntactic and semantic information. The results obtained show promising performances on different tasks when compared to the present state-of-the-art deep learning architectures.

Gpu-based many-core architecture emulation: A double level approach

The efficient emulation of a many-core architecture is a challenging task, each core could be emulated through a dedicated thread and such threads would be interleaved on an either single-core or a multi-core processor. The high number of context switches will results in an unacceptable performance. To support this kind of application, the GPU computational power is exploited in order to schedule the emulation threads on the GPU cores. This presents a non trivial divergence issue, since GPU computational power is offered through SIMD processing elements, that are forced to synchronously execute the same instruction on different memory portions. Thus, a new emulation technique is introduced in order to overcome this limitation: instead of providing a routine for each ISA opcode, the emulator mimics the behavior of the Micro Architecture level, here instructions are date that a unique routine takes as input. Our new technique has been implemented and compared with the classic emulation approach, in order to investigate the chance of a hybrid solution.

Deep learning for computer vision: a comparison between convolutional neural networks and hierarchical temporal memories on object recognition tasks

In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general. However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence". The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them. CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems. HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain. In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.

A deep learning-based approach for 3D people tracking

Questa tesi si occupa dell’estensione di un framework software finalizzato all'individuazione e al tracciamento di persone in una scena ripresa da telecamera stereoscopica. In primo luogo è rimossa la necessità di una calibrazione manuale offline del sistema sfruttando algoritmi che consentono di individuare, a partire da un fotogramma acquisito dalla camera, il piano su cui i soggetti tracciati si muovono. Inoltre, è introdotto un modulo software basato su deep learning con lo scopo di migliorare la precisione del tracciamento. Questo componente, che è in grado di individuare le teste presenti in un fotogramma, consente ridurre i dati analizzati al solo intorno della posizione effettiva di una persona, escludendo oggetti che l’algoritmo di tracciamento sarebbe portato a individuare come persone.

Detecting social patterns within 20th century documentary photos: a deep learning based approach

The job of a historian is to understand what happened in the past, resorting in many cases to written documents as a firsthand source of information. Text, however, does not amount to the only source of knowledge. Pictorial representations, in fact, have also accompanied the main events of the historical timeline. In particular, the opportunity of visually representing circumstances has bloomed since the invention of photography, with the possibility of capturing in real-time the occurrence of a specific events. Thanks to the widespread use of digital technologies (e.g. smartphones and digital cameras), networking capabilities and consequent availability of multimedia content, the academic and industrial research communities have developed artificial intelligence (AI) paradigms with the aim of inferring, transferring and creating new layers of information from images, videos, etc. Now, while AI communities are devoting much of their attention to analyze digital images, from an historical research standpoint more interesting results may be obtained analyzing analog images representing the pre-digital era. Within the aforementioned scenario, the aim of this work is to analyze a collection of analog documentary photographs, building upon state-of-the-art deep learning techniques. In particular, the analysis carried out in this thesis aims at producing two following results: (a) produce the date of an image, and, (b) recognizing its background socio-cultural context,as defined by a group of historical-sociological researchers. Given these premises, the contribution of this work amounts to: (i) the introduction of an historical dataset including images of “Family Album” among all the twentieth century, (ii) the introduction of a new classification task regarding the identification of the socio-cultural context of an image, (iii) the exploitation of different deep learning architectures to perform the image dating and the image socio-cultural context classification.

Previsione del Successo di Prodotti di Moda Prima della Commercializzazione: un Nuovo Dataset e Modello di Vision-Language Transformer

A differenza di quanto avviene nel commercio tradizionale, in quello online il cliente non ha la possibilità di toccare con mano o provare il prodotto. La decisione di acquisto viene maturata in base ai dati messi a disposizione dal venditore attraverso titolo, descrizioni, immagini e alle recensioni di clienti precedenti. É quindi possibile prevedere quanto un prodotto venderà sulla base di queste informazioni. La maggior parte delle soluzioni attualmente presenti in letteratura effettua previsioni basandosi sulle recensioni, oppure analizzando il linguaggio usato nelle descrizioni per capire come questo influenzi le vendite. Le recensioni, tuttavia, non sono informazioni note ai venditori prima della commercializzazione del prodotto; usando solo dati testuali, inoltre, si tralascia l’influenza delle immagini. L'obiettivo di questa tesi è usare modelli di machine learning per prevedere il successo di vendita di un prodotto a partire dalle informazioni disponibili al venditore prima della commercializzazione. Si fa questo introducendo un modello cross-modale basato su Vision-Language Transformer in grado di effettuare classificazione. Un modello di questo tipo può aiutare i venditori a massimizzare il successo di vendita dei prodotti. A causa della mancanza, in letteratura, di dataset contenenti informazioni relative a prodotti venduti online che includono l’indicazione del successo di vendita, il lavoro svolto comprende la realizzazione di un dataset adatto a testare la soluzione sviluppata. Il dataset contiene un elenco di 78300 prodotti di Moda venduti su Amazon, per ognuno dei quali vengono riportate le principali informazioni messe a disposizione dal venditore e una misura di successo sul mercato. Questa viene ricavata a partire dal gradimento espresso dagli acquirenti e dal posizionamento del prodotto in una graduatoria basata sul numero di esemplari venduti.

Analysis of the impact of stationary energy storage systems in trolleybus grids using Simulink-based modelling

The voltage profile of the catenary between traction substations (TSSs) is affected by the trolleybus current intake and by its position with respect to the TSSs: the higher the current requested by the bus and the further the bus from the TSSs, the deeper the voltage drop. When the voltage drops below 500V, the trolleybus is forced to decrease its consumption by reducing its input current. This thesis deals with the analysis of the improvements that the installation of an BESS produces in the operation of a particularly loaded FS of the DC trolleybus network of the city of Bologna. The stationary BESS is charged by the TSSs during off-peak times and delivers the stored energy when the catenary is overloaded alleviating the load on the TSSs and reducing the voltage drops. Only IMC buses are considered in the prospect of a future disposal of all internal combustion engine vehicles. These trolleybuses cause deeper voltage drops because they absorb enough current to power their traction motor and recharge the on board battery. The control of the BESS aims to keep the catenary voltage within the admissible voltage range and makes sure that all physical limitations are met. A model of FS Marconi Trento Trieste is implemented in Simulink environment to simulate its daily operation and compare the behavior of the trolleybus network with and without BESS. From the simulation without BESS, the best location of the energy storage system is deduced, and the battery control is tuned. Furthermore, from the knowledge of the load curve and the battery control trans-characteristic, it is formulated a prediction of the voltage distribution at BESS connection point. The prediction is then compared with the simulation results to validate the Simulink model. The BESS allows to decrease the voltage drops along the catenary, the Joule losses and the current delivered by the TSSs, indicating that the BESS can be a solution to improve the operation of the trolleybus network.

Deep learning based style transfer for low altitude aerial imagery

Unmanned Aerial Vehicle (UAVs) equipped with cameras have been fast deployed to a wide range of applications, such as smart cities, agriculture or search and rescue applications. Even though UAV datasets exist, the amount of open and quality UAV datasets is limited. So far, we want to overcome this lack of high quality annotation data by developing a simulation framework for a parametric generation of synthetic data. The framework accepts input via a serializable format. The input specifies which environment preset is used, the objects to be placed in the environment along with their position and orientation as well as additional information such as object color and size. The result is an environment that is able to produce UAV typical data: RGB image from the UAVs camera, altitude, roll, pitch and yawn of the UAV. Beyond the image generation process, we improve the resulting image data photorealism by using Synthetic-To-Real transfer learning methods. Transfer learning focuses on storing knowledge gained while solving one problem and applying it to a different - although related - problem. This approach has been widely researched in other affine fields and results demonstrate it to be an interesing area to investigate. Since simulated images are easy to create and synthetic-to-real translation has shown good quality results, we are able to generate pseudo-realistic images. Furthermore, object labels are inherently given, so we are capable of extending the already existing UAV datasets with realistic quality images and high resolution meta-data. During the development of this thesis we have been able to produce a result of 68.4% on UAVid. This can be considered a new state-of-art result on this dataset.

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.