Linnaeus University
Not a member yet
43313 research outputs found
Sort by
A Visual Analytics Tool for Discovering Trajectory Patterns Using a Movement Taxonomy
No full textThe analysis of spatio-temporal data has long fascinated researchers, whether by the impact of moving objects based on their trajectory, or by the distinguishing behavioral differences between them. The complexity and heterogeneity of movement data demand considerable effort for meaningful interpretation. To address these concerns, this project proposes a data analytics tool to facilitate our ability to observe and analyze spatio-temporal data. The proposed tool presents a multi-level approach, combining data visualization and statistical computation by categorizing moving objects into distinct taxonomies and implementing Machine Learning models, thereby adding meaningful structure to the analysis. Two case studies were performed as part of the methodological approach. The first case study was performed using a dataset of Arctic foxes' trajectories. Foxes were successfully identified and labeled as having Geometric or Kinematic-based behavior, and then further categorized into Curvature and Acceleration groups, identifying unique statistical indicators that represented such behaviors. A pattern was discovered, which showed that foxes with Acceleration-based behavior presented constant and steady acceleration. Those represented instead by Curvature had acceleration peaks and sudden deceleration throughout their trajectory paths. The second case study analyzed tropical cyclone data. The tool was able to label the trajectories with Speed, Curvature, and hybrid Geometric-based behaviors, finding unique statistical variables representing them. A deeper analysis evaluating individual trajectories with hybrid Geometric behavior (i.e., Curvature and Indentation combined) explicitly identified the unique angles with the highest impact on a hurricane's shape and geometry. The effectiveness of the method and tool proposed shows that spatio-temporal data, despite its inherent complexity, can still be analyzed and explained in detail, and provides a theoretical and practical blueprint that may be applied to several application areas
Efficient Production in the Wood Processing Industry : Lean Principles and AI-Based Image Analysis in the Cutting Process
No full textBakgrund: Sveriges omfattande skogsresurser har möjliggjort utvecklingen av en teknologiskt avancerad träbearbetningsindustri, där inkapningsprocessen utgör en kritisk del av värdekedjan. I takt med en globalt ökad efterfrågan på högkvalitativa och hållbara träprodukter, ställs företag inför krav på att optimera produktionseffektivitet och minska materialspill. Trämaterialets oförutsägbara natur skapar höga krav på både standardisering och flexibilitet i processerna. Lean-principer, med fokus på att minska slöseri och förutsägbarhet, tillämpas ofta inom tillverkning för att uppnå effektivitet. Samtidigt har AI-baserad bildanalys, särskilt genom tvådimensionella kameratekniker, etablerat sig som ett digitalt verktyg för automatiserad kvalitetskontroll och beslutsstöd. Trots att Lean och AI var för sig bidragit till förbättrad produktionseffektivitet, saknas fortfarande forskning som kombinerar de tillvägagångssätten inom träindustrin. Studien behandlar den forskningsluckan genom att undersöka hur Lean-principer och AI-baserad bildanalys kan integreras för att förbättra standardisering och effektivitet i inkapningsprocessen. Syfte: Syftet med studien är att identifiera och minska slöseri enligt Lean-principer samt att undersöka hur AI kan stödja beslutsfattande och resursanvändning i en oförutsägbar produktionsmiljö. Studien fokuserar även på att analysera hur Lean och AI-baserad bildanalys kan samverka för att standardisera och effektivisera inkapningsprocessen hos Företag X. Metod: Studien har genomförts som en kvalitativ fallstudie med ett kritiskt realistiskt vetenskapligt synsätt och ett deduktivt angreppssätt. Empirin baseras på semistrukturerade intervjuer, icke-deltagande observationer och dokumentanalys, vilket möjliggör datatriangulering och kontextuell förståelse. Urvalet har skett genom målstyrt och snöbollsurval för att identifiera nyckelpersoner inom och omkring inkapningsprocessen. Den insamlade datan har analyserats med tematisk analys utifrån de teoretiska ramverken Lean och AI-baserad bildanalys. Kodningen strukturerades utifrån Leans åtta slöserier, vilket möjliggjorde en systematisk tolkning av hur varje slöseri yttrar sig i processen. Etiska principer såsom konfidentialitet, anonymitet och informerat samtycke har tillämpats under hela forskningsprocessen. Resultat: Resultatet visar att samtliga åtta slöserier enligt Lean med hjälp av VSM, nämligen väntan, onödiga transporter, överarbete, lager, onödiga rörelser, defekter, outnyttjad kompetens samt i viss mån överproduktion, förekommer i Företag X:s inkapningsprocess. Genom en framtida värdeflödeskartläggning (FVSM) kopplades förbättringsåtgärder till digitalisering, standardisering, operatörsdelaktighet samt strukturerad lagerstyrning enligt FIFO, vilket syftar till att minska ledtider, förbättra resursutnyttjande och öka hållbarheten. AI-baserad bildanalys påverkar det fysiska flödet genom att möjliggöra realtidsåterkoppling, standardiserad sortering och datadrivet beslutsfattande, men tekniken har begränsningar såsom beroende av omfattande träningsdata och svårighet att upptäcka djupliggande defekter. De faktorerna kräver fortsatt samverkan mellan människa och maskin. En parallell tillämpning av Lean och AI-baserad bildanalys stärker processen, där Lean skapar strukturella förutsättningar och AI bidrar med datadrivet förbättringsarbete. Tillsammans främjar de standardisering och effektivisering även i en miljö präglad av materialets oförutsägbarhet.Background: Sweden's extensive forest resources have enabled the development of a technologically advanced wood processing industry, where the cutting process represents a critical part of the value chain. As global demand for high quality, sustainable wood products increases, companies are pressured to optimize production efficiency and reduce material waste. The unpredictable nature of wood as a raw material places high demands on both standardization and process flexibility. Lean principles, emphasizing waste reduction and process predictability, are commonly applied in manufacturing to achieve efficiency. At the same time, AI-based image analysis, particularly through two-dimensional vision systems, has emerged as a digital tool for automating quality control and decision-making. Although Lean and AI have individually contributed to performance improvements, research combining these two approaches in wood processing remains limited. This study addresses that gap by exploring how Lean principles and AI-based image analysis can be integrated to improve standardization and efficiency in the cutting process. Purpose: The purpose of this study is to identify and reduce waste based on Lean principles, and to examine how AI can support decision making and resource utilization in an unpredictable production environment. The study also focuses on analyzing how Lean and AI-based image analysis can interact to standardize and improve the efficiency of the cutting process at Company X. Method: This study adopts a qualitative single case study design grounded in critical realism and a deductive research approach. The empirical foundation consists of semi-structured interviews, non-participant observations, and document analysis, allowing for data triangulation and contextual understanding. A purposive and snowball sampling strategy was used to identify key informants involved in the cutting process and its surrounding activities. The data were analyzed using thematic analysis, guided by theoretical frameworks on Lean principles and AI-based image analysis. The coding process was structured around the eight forms of waste identified in Lean, enabling a systematic interpretation of how each type of waste manifests within the cutting process. Ethical considerations, including confidentiality, anonymity, and informed consent, were strictly observed throughout the research process. Results: The results show that all eight forms of waste defined by Lean with the help of VSM, waiting, unnecessary transport, overprocessing, inventory, unnecessary motion, defects, underutilized talent, and to some extent overproduction, are present in Company X’s cutting process. Through a future value stream mapping (FVSM), improvement opportunities were linked to digitalization, standardization, operator involvement, and structured inventory control (FIFO), aiming to enhance lead times, resource utilization, and sustainability. AI-based image analysis influences the physical flow by enabling real-time feedback, standardization in sorting, and data driven decisions, yet its limitations, such as dependency on training data and difficulty detecting internal defects, require continued human–machine collaboration. When applied in parallel, Lean provides structural stability while AI strengthens continuous improvement through analytical decision support, together promoting process standardization and efficiency in the face of raw material unpredictability
Testbed for photovoltaic regulator data transmission
No full textThe integration of photovoltaic (PV) systems into decentralized energy networks requires reliable data transmission from solar regulators and microinverters. This project investigates the data transmission capabilities of various PV regulators and inverters, focusing on their communication protocols and performance metrics. The primary problem addressed is the lack of comprehensive documentation on data transmission speeds from these devices. By creating a controlled testbed, we systematically evaluate the data transmission intervals, current detection speed, and reliability of different regulators and inverters. Our results provide insights into the limitations and capabilities of these devices and contribute valuable information for the development of efficient decentralized energy systems
Representationer av blodomloppet : Undersökning av tre läroböcker i biologi årskurs 4-6
No full textSyftet med studien är att analysera hur blodomloppet presenteras i läroböcker i mellanstadiet. Metoden som använts är en kvalitativ läroboksanalys med tre läroböcker som finns i ämnet biologi. I de tre läroböckerna lokaliserades 16 representationer av blodomloppet som sedan analyserades. Resultatet av representationer i läroböckerna har analyserats med hjälp av fyra olika kriterier som beskriver vad som innefattas i representationerna. Resultatet visar att majoriteten av representationerna innehöll multipla representationsnivåer. Det visade också att ytegenskaperna var övervägande explicita vilket innebar att de förklarade samtliga viktiga beståndsdelar i representationen. Slutligen visade resultatet att representationerna hade oftast ett delvis samband till texten och även bildtexter som klargjorde det övergripande fenomenet av representationen. Tidigare forskning beskriver blodomloppet som ett abstrakt och komplext system som samverkar med många andra system i människokroppen. Resultaten indikerar att läroböckerna består av representationer om kan främja elevernas lärande inom biologin.
Application of SAM2 for Defect Detection in Manufacturing
No full textUtilizing computer vision to automate defect inspection in manufacturing requires large, accurately annotated datasets to train effective models. Manual annotation of these datasets is a time-consuming and costly process for industry stakeholders. This thesis investigates the use of Meta’s open-source Segment Anything Model 2 (SAM2) to streamline and reduce the cost of dataset annotation in this domain. SAM2’s segmentation performance is evaluated against a real-world dataset containing manual annotations provided by industry, serving as a benchmark for comparison. We also explore how model fine-tuning and image augmentation affect segmentation quality. Our results suggest that SAM2 has the potential to be used as a semi-automated data set annotation tool, although more research is needed to fully assess its scalability and effectiveness in industrial workflows
So you are carrying
No full textDetta teaterstycke börjar med att en kvinna i rullstol, fru A, sitter och skalar ett äpple med en äppelkniv. En man, Herr B, kommer in från vänster på scen och bär på en väska, varpå fru A konstaterar: ”Du bär du”. De har aldrig träffats förut och snart uppstår en konversation mellan dem som handlar om vad som finns i Herr Bs väska och vart han är på väg med den. Den frågan återkommer genom hela pjäsen. Även äpplet och äppelkniven blir också återkommande element i pjäsen. Det uppstår både missförstånd och visst samförstånd mellan Fru A och Herr B under pjäsens gång. Ett par, fröken C och herr D, dyker upp ganska tidigt i pjäsen. De är på resande fot med varsin rullbar kabinväska. Deras konversation liksom de delar av deras samliv vi får del av, består också av ett visst mått av samförstånd men ett stort mått av missförstånd. En viktig roll i pjäsen spelar också de mobiltelefoner som alla bär på och som leder dem på deras väg, men också fungerar som instrument för att misskreditera kontrahenterna. Pjäsen är skriven utifrån en vilja att låta författarens intuition vara vägledande för vad som utspelar sig på scen, snarare än att ha ett färdigt synopsis innan skrivprocessen inletts. Scener och repliker har dock efterredigerats så att det som sker på scenen ska vara förståeligt och ha en viss konsekvens. Pjäsen uppvisar vissa likheter med den absurda teatertraditionen.
Political Participation and Engagement Among Youth : A Qualitative Study on Political Engagement, Participation, and Trust Among Young People
No full textThis study examines youth political engagement and the multiple factors shaping their participation. It aims to deepen the understanding of young individuals' relationship with politics by exploring how trust, political confidence, and socioeconomic conditions influence their engagement. The study is guided by three core questions: how youth political engagement is changing, how it is expressed, and how factors such as socioeconomic background, gender, age, and cultural capital affect political socialization and participation. Using a qualitative approach, the research analyzes interviews with 16 young individuals, structuring findings thematically around engagement, trust, forms of participation, and emotional-political expression. The study draws upon sociological theories, including social capital, cultural capital and habitus (Bourdieu). Findings reveal a generational shift in political engagement, where many young people rely on informal or digital forms of activism, such as social media and petitions. While interest in politics remains high, traditional participation is often hindered by low trust in institutions, perceived complexity, or lack of emotional connection. The way youths speak about politics, marked by insecurity, cynicism, or confident idealism, is closely tied to their access to political discourse in the home, their social environment, and their symbolic and linguistic capital. The study concludes that understanding youth political engagement requires not only acknowledging evolving modes of participation but also recognizing how social and emotional factors shape their access, trust, and motivation to act
The work against homelessness in small municipalities : A qualitative interview study about Housing First and its implementation
No full textHousing First is a model against homelessness that has existed in Sweden since 2011. Furthermore, since 2022, there is a national strategy for working against homelessness, including an ambition to introduce Housing First in all the swedish municipalities. However, out of all the 134 municipalities that are categorised as small in Sweden, only 38 of them has chosen to implement the method by now, and there is less research on this specific topic of implementation of Housing first in small municipalities. Therefore this study aims to explore the implementation process of Housing First in small Swedish municipalities, and investigate what possibilities and challenges they have met while implementing the method. Data for this study has been collected through qualitative interviews with social workers from small municipalities. The data was analyzed by a thematic analysis with Lipsky's theory of Street-level Bureaucrats and Nida's translation theory as theoretical framework. The analysis showed that it is hard for the small municipalities to work faithfully to the method. It also showed that relations, commitment and cooperation between the social workers and external actors were important for a successful implementation. The result also showed several challenges when implementing the method. Challenges were the mindset of the surroundings, lack of accommodations and the landlords high requirements.
Women and Men in Social Work : A Qualitative Study of the Journal <em data-start="127" data-end="139">Socionomen
No full textWomen have long held a central role in social work in Sweden, where their traditionally caring qualities have been associated with femininity. Historically, they worked on a voluntary basis, but gained greater access to the labor market as health and social care became professionalized. Despite this progress, women still take on a larger share of domestic responsibilities and face greater challenges in reaching leadership positions, which is linked to prevailing gender norms. Researchers argue that women are socialized into caring roles, while men are encouraged toward independence and leadership—norms that continue to influence career choices and power structures in society. We have collected our data from opinion articles published in the journal Socionomen and analyzed them through an interpretive content analysis using Yvonne Hirdman’s theory of gender systems. The study shows that male and female writers in the journal Socionomen tend to choose different topics and forms of expression, with men more frequently writing with agency and women with care. This division reflects and reproduces prevailing gender norms within social work and society at large. The results indicate how traditional gendered assumptions continue to shape which positions and modes of expression are considered legitimate. By making this pattern visible, we highlight the need for greater awareness of the impact of gender norms, both in education and professional practice. The study thus contributes to an understanding of how gender inequality can be reinforced even within female-dominated fields
Enabling workplace learning : A study of the interplay between individuals, colleagues and leadership
No full textSyftet med denna studie är att undersöka medarbetares upplevelser av lärande i arbetslivet samt vilka faktorer som främjar och hindrar deras lärandeprocesser. Studien grundas i en kvalitativ metod med semistrukturerade intervjuer och tematisk analys. Det teoretiska perspektivet som använts i studien är Lave och Wengers (1991) Situated Learning Theory, som betonar lärande som en social och kontextbunden process. Resultatet visar att kollegialt samspel, motivation och ledarskapsstöd är centrala faktorer för lärande i arbetet. Samtidigt identifieras faktorer som brist på tid och resurser som betydande hinder för lärandet. Resultatet visar även att ett stödjande ledarskap med fokus på delaktighet och ansvar främjar lärandemiljön. Slutsatsen är att lärande på arbetsplatsen sker genom ett samspel mellan individ, kollegor och organisatoriska strukturer, där informellt lärande spelar en avgörande roll