This thesis describes the design and development of a system for detection and recognition of license plates. The work is divided into three basic parts: licence plates detection, finding of character positions and optical character recognition. To fullfill the goal of this work, a new dataset was taken. It contains 2814 license plates used for training classifiers and 2620 plates to evaluate the success rate of the system. Cascade Classifier was used to train detector of licence plates, which has success rate up to 97.8 %. After that, pozitions of individual characters were searched in detected pozitions of licence plates. If there was no character found, detected pozition was not the licence plate. Success rate of licence plates detection with all the characters found is up to 88.5 %. Character recognition is performed by SVM classifier. The system detects successfully with no errors up to 97.7 % of all licence plates.

Orientador: João Antonio Pereira / Resumo: Visão de máquina é um campo multidisciplinar que vem crescendo na indústria, que está cada vez mais preocupada em reduzir custos, automatizar processos, e atender requisitos de qualidade do produto para atender seus clientes. Processos de montagem realizados de forma manual com inspeção e controle visual são tipicamente processos susceptíveis a erros, à utilização de peças não conformes na montagem do produto final. Este trabalho apresenta uma proposta de desenvolvimento de um sistema de visão de máquina com base no processamento e análise de imagens digitais para a inspeção das características e especificações das peças e componentes utilizados na montagem de capotas marítimas, objetivando verificar e garantir a conformidade do produto final. A inspeção e avaliação da conformidade do produto são feitas por etapas com a utilização de duas câmeras, uma captura a imagem do código de identificação alfanumérico do produto e a outra inspeciona o conjunto de elementos de fixação. As imagens passam por um processo de tratamento que envolve a filtragem espacial utilizando máscara de médias para suavização, alargamento de contraste para expandir a faixa de intensidades e segmentação para formação dos objetos de interesse. Uma função de OCR é utilizada para a extração de caracteres e reconhecimento do código do produto e a extração de características específicas do conjunto de componentes de fixação é feita por descritores de forma representados pelos invariantes de momento. As caracte... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Machine vision is a growing multidisciplinary field in the industry that is increasingly concerned with reducing costs, automating processes, and meeting product quality requirements to serve its customers. Manual assembly processes with inspection and visual control are typically error-prone processes using non-conforming parts in the final product assembly. This work presents a proposal for the development of a machine vision system based on digital image processing and analysis for the inspection of the characteristics and specifications of the parts and components used in the assembly of marine bonnets, aiming to verify and ensure the conformity of the final product. Inspection and conformity assessment of the product are done in stages using two cameras, one capturing the image of the alphanumeric identification code of the product and the other inspecting the set of fasteners. The images undergo a treatment process that involves spatial filtering using averaging masks for smoothing, contrast widening to expand the range of intensities, and segmentation to form the objects of interest. An OCR function is used for character extraction and product code recognition, and the extraction of specific features of the fastener assembly is done by shape descriptors represented by the moment invariants. The specific characteristics of the fasteners are used to assess the conformity of the product with its respective code. The presentation of data and results of the implemented prop... (Complete abstract click electronic access below) / Mestre

This thesis explores the implementation of Optical Character Recognition (OCR) – based text extraction and random forest regression analysis for housing market valuation, specifically focusing on the impact of value factors, derived from OCR-extracted economic values from housing cooperatives’ annual reports. The objective is to perform price estimations using the Random Forest model to identify the key value factors that influence the estimation process and examine how the economic values from annual reports affect the sales price. The thesis aims to highlight the often-overlooked aspect that when purchasing an apartment, one also assumes the liabilities of the housing cooperative. The motivation for utilizing OCR techniques stems from the difficulties associated with manual data collection, as there is a lack of readily accessible structured data on the subject, emphasizing the importance of automation for effective data extraction. The findings indicate that OCR can effectively extract data from annual reports, but with limitations due to variation in report structures. The regression analysis reveals the Random Forest model’s effectiveness in estimating prices, with location and construction year emerging as the most influential factors. Furthermore, incorporating the economic values from the annual reports enhances the accuracy of price estimation compared to the model that excluded such factors. However, definitive conclusions regarding the precise impact of these economic factors could not be drawn due to limited geographical spread of data points and potential hidden value factors. The study concludes that the machine learning model can be used to make a credible price estimate on cooperative apartments and that OCR methods prove valuable in automating data extraction from annual reports, although standardising report format would enhance their efficiency. The thesis highlights the significance of considering the housing cooperatives’ economic values when making property purchases.

Dysgraphia is a rather widespread learning disorder in the current society. It is well established that an early diagnosis of this writing disorder can lead to improvement in writing skills. However, as of today, although there is no comprehensive standard process for the evaluation of dysgraphia, most of the tests used for this purpose must be done at a physician’s office. On the other hand, the pandemic triggered by COVID-19 has forced people to stay at home and opened the door to the development of online medical consultations. The present study therefore aims to propose an automated pipeline to provide pre-clinical diagnosis of dysgraphia. In particular, it investigates the possibility of applying deep learning techniques to the most widely used test for assessing writing difficulties in Italy, the BVSCO-2. This test consists of several writing exercises to be performed by the child on paper under the supervision of a doctor. To test the hypothesis that it is possible to enable children to have their writing impairment recognized even at a distance, an innovative system has been developed. It leverages an already developed customized tablet application that captures the graphemes produced by the child and an artificial neural network that processes the images and recognizes the handwritten text. The experimental results were analyzed using different methods and were compared with the actual diagnosis that a doctor would have provided if the test had been carried out normally. It turned out that, despite a slight fixed bias introduced by the machine for some specific exercises, these results seemed very promising in terms of both handwritten text recognition and diagnosis of children with dysgraphia, thus giving a satisfactory answer to the proposed research question. / Dysgrafi är en ganska utbredd inlärningsstörning i dagens samhälle. Det är väl etablerat att en tidig diagnos av denna skrivstörning kan leda till en förbättring av skrivförmågan. Även om det i dag inte finns någon omfattande standardprocess för utvärdering av dysgrafi måste dock de flesta av de tester som används för detta ändamål göras på en läkarmottagning. Å andra sidan har den pandemi som utlöstes av COVID-19 tvingat människor att stanna hemma och öppnat dörren för utvecklingen av medicinska konsultationer online. Syftet med denna studie är därför att föreslå en automatiserad pipeline för att ge preklinisk diagnos av dysgrafi. I synnerhet undersöks möjligheten att tillämpa djupinlärningstekniker på det mest använda testet för att bedöma skrivsvårigheter i Italien, BVSCO-2. Testet består av flera skrivövningar som barnet ska utföra på papper under överinseende av en läkare. För att testa hypotesen att det är möjligt att göra det möjligt för barn att få sina skrivsvårigheter erkända även på distans har ett innovativt system utvecklats. Det utnyttjar en redan utvecklad skräddarsydd applikation för surfplattor som fångar de grafem som barnet producerar och ett artificiellt neuralt nätverk som bearbetar bilderna och känner igen den handskrivna texten. De experimentella resultaten analyserades med hjälp av olika metoder och jämfördes med den faktiska diagnos som en läkare skulle ha ställt om testet hade utförts normalt. Det visade sig att, trots en liten fast bias som maskinen införde för vissa specifika övningar, verkade dessa resultat mycket lovande när det gäller både igenkänning av handskriven text och diagnos av barn med dysgrafi, vilket gav ett tillfredsställande svar på den föreslagna forskningsfrågan.

<p dir="ltr">Building Information Modeling (BIM) supports engineering and performance analysis for buildings and civil infrastructure from the initial design stage. BIM offers engineers access to building and infrastructure objects, along with their associated data, which can be utilized across various platforms to develop analytical models. However, the interoperability between BIM and analytical models is still limited and challenging. One such limitation and challenge is in the interoperability between BIM and Building Energy Modeling (BEM). Despite the fact that interoperability of geometry and material information between BIM and BEM has been extensively investigated, the interoperability of heating, ventilation, and air conditioning (HVAC) information, which is a crucial part in BEM, was underinvestigated. Another limitation is that the shared objects frequently lose their identification across different models during the processes of their creation, design iterations, and model transformation. In addition, current building and civil infrastructure projects mainly rely on Portable Document Format (PDF) plans as the official deliverables and documents to be stored, communicated, and transferred among different stakeholders. The transition from 2D PDF plans to 3D BIM remains challenging because manually creating a BIM instance model from 2D drawings can be laborious, time-intensive, and susceptible to errors.</p><p dir="ltr">To address these gaps, this dissertational research introduces new Industry Foundation Classes (IFC)-based algorithmic methods that utilize the state-of-the-art Data-driven Reverse Engineering Algorithm Development (D-READ) method and the invariant signatures of architecture, engineering, and construction (AEC) objects to (1) develop algorithms that can extract the information from 2D PDF drawings and reconstruct the 3D semantically segmented and enriched BIM instance models, (2) develop object mapping algorithms for interfacing BIM and analytical models (e.g., BEM, structural analysis models, etc.) by automatically mapping building objects, and (3) iteratively develop the HVAC information transformation algorithm between BIM and BEM. Following the proposed methods, algorithms were developed to (1) semi-automate the creation of semantically segmented and enriched 3D IFC-based bridge BIM instance models using 2D PDF bridge plans, (2) map space objects between BIM instance models and BEM (OpenStudio model) based on their invariant signatures, and (3) transform HVAC objects from IFC-based BIM instance models to BEM with all the necessary information for energy simulation, using (1) PDF drawings for 12 bridges located in various parts of Indiana, (2) a 2-story duplex apartment building, and (3) a 2-story office building model and a 2-story residential building model, respectively.</p><p dir="ltr">The developed algorithms were tested on three cases: (1) the PDF information extraction algorithm was tested on six bridges, which achieved 97.7% precision and 94.4% recall. In addition, it decreased the time required to create bridge BIM instance models by 94.9% compared to the manual approach; (2) the object mapping algorithm was evaluated using a 4-story office building model containing 82 spaces. The results demonstrated that the algorithm attained 90% precision and 90% recall in mapping space objects. Additionally, a 4.88% improvement in the accuracy of energy simulation results was observed when compared to simulations without space mapping; (3) the HVAC transformation algorithm was tested on two models with distinct HVAC systems: a 4-story office building model featuring a boiler radiator system and a 2-story clinic building featuring a VAV system. The algorithm achieved transformation accuracies of 97.5% and 98.7%, respectively, compared to manually created evaluation models in OpenStudio. Additionally, the algorithm-generated models demonstrated satisfactory performance with regard to precision, with less than 9.6% error in total annual energy consumption compared to the evaluation models.</p><p dir="ltr">This dissertational research introduces a new IFC-based approach to fill the forementioned research gaps in BIM interoperability for buildings and civil infrastructures. It facilitates improved accessibility compared to a proprietary workflow and will contribute to filling informational gaps (1) between 3D BIM and 2D PDF drawings, and (2) between BIM and analytical models. It builds a solid foundation for achieving (1) automated BIM reconstruction using 2D plans, and (2) smooth, accurate, and fully-automated HVAC objects transformation between BIM and BEM, for complete BIM-BEM interoperability. The proposed approach can also be leveraged to further expand BIM interoperability support by providing a novel data-driven approach for building and civil infrastructure projects.</p>

Les travaux de recherche présentés dans ce manuscrit décrivent plusieurs apports au thème de l’évaluation de la qualité d’images de documents numérisés. Pour cela nous proposons de nouveaux descripteurs permettant de quantifier les dégradations les plus couramment rencontrées sur les images de documents numérisés. Nous proposons également une méthodologie s’appuyant sur le calcul de ces descripteurs et permettant de prédire les performances d’algorithmes de traitement et d’analyse d’images de documents. Les descripteurs sont définis en analysant l’influence des dégradations sur les performances de différents algorithmes, puis utilisés pour créer des modèles de prédiction à l’aide de régresseurs statistiques. La pertinence, des descripteurs proposés et de la méthodologie de prédiction, est validée de plusieurs façons. Premièrement, par la prédiction des performances de onze algorithmes de binarisation. Deuxièmement par la création d’un processus automatique de sélection de l’algorithme de binarisation le plus performant pour chaque image. Puis pour finir, par la prédiction des performances de deux OCRs en fonction de l’importance du défaut de transparence (diffusion de l’encre du recto sur le verso d’un document). Ce travail sur la prédiction des performances d’algorithmes est aussi l’occasion d’aborder les problèmes scientifiques liés à la création de vérités-terrains et d’évaluation de performances. / This PhD. thesis deals with quality evaluation of digitized document images. In order to measure the quality of a document image, we propose to create new features dedicated to the characterization of most commons degradations. We also propose to use these features to create prediction models able to predict the performances of different types of document analysis algorithms. The features are defined by analyzing the impact of a specific degradation on the results of an algorithm and then used to create statistical regressors.The relevance of the proposed features and predictions models, is analyzed in several experimentations. The first one aims to predict the performance of different binarization methods. The second experiment aims to create an automatic procedure able to select the best binarization method for each image. At last, the third experiment aims to create a prediction model for two commonly used OCRs. This work on performance prediction algorithms is also an opportunity to discuss the scientific problems of creating ground-truth for performance evaluation.

This thesis describes a development of a modern client-server application for text recognition on iOS platform. The reader is acquainted with common principles of a client-server model, including its known architecture styles, and with a distribution of logical layers between both sides of the model. After that the thesis depicts current trends and examples of suitable technologies for creating an application programming interface of a web server. Possible ways of text recognition on the server side are discussed as well. In context of a client side, the thesis provides an insight into iOS platform and a few important concepts in iOS application development. Following implementation of the server side is stressed to be reusable as much as possible for different kinds of use cases. Last but not least, the thesis provides a simple iOS framework for a direct communication with the recognition server. Finally, an application for evaluation of food ingredients from a packaging material is implemented as an example of usage.

This report explores the implementation of AI techniques in the manufacturing process at Ovako, focusing on process optimization, individual traceability, and quality control. By integrating advanced AI models and techniques at various levels within the production process, Ovako can improve efficiency, reduce material consumption, and prevent production stops. For example, predictive maintenance can be applied to anticipate and prevent machine problems, while image recognition algorithms and optical character recognition enable individual traceability of each rod throughout the process. Furthermore, AI-based quality control can detect defects and deviations with high precision and speed, leading to reduced risk of faulty products and increased product quality. By carefully considering the role of the workforce, safety and ethical issues, and the benefits and challenges of AI implementation, Ovako can maximize the benefits of these techniques and enhance its competitiveness in the market. / Denna rapport utforskar implementeringen av AI-tekniker i tillverkningsprocessen hos Ovako, med fokus på processoptimering, individuell spårbarhet och kvalitetskontroll. Genom att integrera avancerade AI-modeller och tekniker på olika nivåer inom produktionsprocessen kan Ovako förbättra effektiviteten, minska materialförbrukningen och förhindra produktionsstopp. Exempelvis kan prediktivt underhåll tillämpas för att förutse och förebygga maskinproblem, medan bildigenkänningsalgoritmer och optisk teckenigenkänning möjliggör individuell spårbarhet av varje stång genom processen. Dessutom kan AI-baserad kvalitetskontroll detektera defekter och avvikelser med hög precision och hastighet, vilket leder till minskad risk för felaktiga produkter och ökad produktkvalitet. Genom att noggrant överväga arbetskraftens roll, säkerhets- och etikfrågor samt fördelarna och utmaningarna med AI-implementeringen kan Ovako maximera nyttan av dessa tekniker och förbättra sin konkurrenskraft på marknaden.