Bag-of-Features Approach to Unsupervised Visual Object Categorisation

The large and growing number of digital images is making manual image search laborious. Only a fraction of the images contain metadata that can be used to search for a particular type of image. Thus, the main research question of this thesis is whether it is possible to learn visual object categories directly from images. Computers process images as long lists of pixels that do not have a clear connection to high-level semantics which could be used in the image search. There are various methods introduced in the literature to extract low-level image features and also approaches to connect these low-level features with high-level semantics. One of these approaches is called Bag-of-Features which is studied in the thesis. In the Bag-of-Features approach, the images are described using a visual codebook. The codebook is built from the descriptions of the image patches using clustering. The images are described by matching descriptions of image patches with the visual codebook and computing the number of matches for each code. In this thesis, unsupervised visual object categorisation using the Bag-of-Features approach is studied. The goal is to find groups of similar images, e.g., images that contain an object from the same category. The standard Bag-of-Features approach is improved by using spatial information and visual saliency. It was found that the performance of the visual object categorisation can be improved by using spatial information of local features to verify the matches. However, this process is computationally heavy, and thus, the number of images must be limited in the spatial matching, for example, by using the Bag-of-Features method as in this study. Different approaches for saliency detection are studied and a new method based on the Hessian-Affine local feature detector is proposed. The new method achieves comparable results with current state-of-the-art. The visual object categorisation performance was improved by using foreground segmentation based on saliency information, especially when the background could be considered as clutter.

Alteração do campo visual em função do uso de cabinas em tratores agrícolas

Com o objetivo de avaliar a alteração do campo visual de um trator, ocasionada pela instalação de uma cabina, foram determinadas as áreas de visibilidade nula em função da presença ou não da mesma, para cada campo visual. Os procedimentos adotados basearam-se na norma ISO 5721-1981 - Agricultural tractors - Operator's field of vision. A cabina aumentou em 34,42% a área de visibilidade nula. A relação área de projeção do trator/área total sombreada foi de 0,075 para o trator sem cabina, e 0,057 para o trator com cabina, indicando as melhores condições de visibilidade do primeiro em relação ao segundo. No que se refere aos campos visuais, a maior redução ocorreu no campo visual dos olhos e no obtido com o movimento da cabeça, em função dos suportes da cabina e da estrutura de proteção contra o capotamento. A cabina diminui o campo de visão do operador, e adequações no projeto da estrutura avaliada são necessárias para melhorar as condições de visibilidade.

Avaliação de métodos alternativos à citologia no rastreamento de lesões cervicais: detecção de DNA-HPV e inspeção visual

OBJETIVO: avaliar o desempenho da citologia oncológica (CO), de captura híbrida II (CH II) e da inspeção visual com ácido acético na detecção de lesões pré-neoplásicas e neoplásicas cervicais. MÉTODOS: 2281 mulheres foram submetidas a exame clínico além da coleta de material para citologia, para CH II para detecção de DNA-HPV. Foi feita a inspeção visual do colo após aplicação do ácido acético a 5% (IVA). As mulheres com pelo menos um exame positivo eram convocadas para colposcopia, que também foi realizada em 420 mulheres com todos os exames normais. O desempenho dos testes foi calculado utilizando como padrão ouro a colposcopia com ou sem biopsia. RESULTADOS: a CO foi anormal em 209 mulheres (9,2%); a CH II foi positiva em 399 (17,5%) e em 249 (10,9%) foram encontradas alterações na IVA. Entre as 2281 mulheres avaliadas, 671 (29,4%) apresentaram pelo menos um resultado de exame positivo, embora apenas 82 (3,6%) apresentassem doença confirmada histologicamente (50 NIC1, 20 NIC2, sete NIC3 e cinco carcinomas invasores). As sensibilidades da IVA e da CH II foram semelhantes e significativamente maiores que a da CO. A especificidade da CO foi maior que a da IVA e da CH II. Nos casos com resultado de CO negativo, a IVA apresentou melhor desempenho comparada à CH II. CONCLUSÃO: o desempenho da CO associada à IVA foi melhor que o da CO associada à CH II e do que o da CO isolada.

Inspeção visual do colo uterino após aplicação de ácido acético no rastreamento das neoplasias intra-epiteliais e lesões induzidas por HPV

OBJETIVO: estimar a validade da inspeção visual após aplicação de ácido acético (IVA) no rastreamento das neoplasias intra-epiteliais cervicais (NIC) e lesões induzidas por HPV, comparando seu desempenho com o da colpocitologia e da colposcopia. MÉTODOS: estudo de validação de teste diagnóstico realizado em 893 mulheres de 18 a 65 anos, rastreadas simultaneamente com colpocitologia, IVA e colposcopia, em unidade de saúde pública de Recife, PE. A IVA foi realizada por embrocação do colo com ácido acético a 5% e observação a olho nu, com auxílio de foco clínico comum. Considerou-se como positividade o achado de qualquer lesão aceto-branca no colo. O padrão-ouro foi o histopatológico de biópsia cervical, realizado sempre que qualquer um dos três testes resultasse anormal. Foram estimados e comparados os indicadores de validade de cada teste, com os respectivos intervalos de confiança a 95%. A concordância entre os resultados dos testes foi avaliada pelo coeficiente kappa (k). RESULTADOS: das 303 mulheres biopsiadas, o estudo histopatológico foi anormal em 24. Deste total, a IVA foi positiva em 22, conferindo-lhe sensibilidade estimada de 91,7%, especificidade de 68,9%, valor preditivo positivo de 7,5% e valor preditivo negativo de 99,7%. Comparando-se os intervalos de confiança a 95%, a IVA mostrou maior sensibilidade que a colpocitologia, mas com menores especificidade e valor preditivo positivo. Houve fraca concordância entre os resultados da IVA e da colpocitologia (k=0,02) e excelente concordância com os da colposcopia (k=0,93). CONCLUSÃO: a IVA foi muito mais sensível que a colpocitologia no rastreamento das NIC e lesões HPV-induzidas e teve o mesmo desempenho da colposcopia. Sua baixa especificidade foi responsável por um elevado número de resultados falso-positivos.

Lean daily management, visual management and continuous improvement

This thesis was part of lean adaptation project started at Outotec Lappeenranta factory in early 2013. The purpose of this thesis was to develop and propose lean tools that could be used in daily management, visual management and continuous improvement. This thesis was “outsiders” view, and as such, did not study the current processes deeply. As result of this thesis, two different Daily Management -boards were designed, one for parallel processes and one for sequential processes. In addition, methods of doing continuous improvement and daily task accountability were framed and standard work for the leaders outlined. The tools presented in this thesis are general tools which support work in lean environment. They are visual and, if used correctly, they provide a basis from which continuous improvement can be done. Lean philosophy emphasizes the deep understanding of the current situation and it would be against the lean principles to blindly implement anything developed “on the outside”. The tools presented should be reviewed and modified further by the people working on the factory floor.

Visual customer segmentation and behavior analysis : a SOM-based approach

Ett ämne som väckt intresse både inom industrin och forskningen är hantering av kundförhållanden (CRM, eng. Customer Relationship Management), dvs. en kundorienterad affärsstrategi där företagen från att ha varit produktorienterade väljer att bli mera kundcentrerade. Numera kan kundernas beteende och aktiviteter lätt registreras och sparas med hjälp av integrerade affärssystem (ERP, eng. Enterprise Resource Planning) och datalager (DW, eng. Data Warehousing). Kunder med olika preferenser och köpbeteende skapar sin egen ”signatur” i synnerhet via användningen av kundkort, vilket möjliggör mångsidig modellering av kundernas köpbeteende. För att få en översikt av kundernas köpbeteende och deras lönsamhet, används ofta kundsegmentering som en metod för att indela kunderna i grupper utgående från deras likheter. De mest använda metoderna för kundsegmentering är analytiska modeller konstruerade för en viss tidsperiod. Dessa modeller beaktar inte att kundernas beteende kan förändras med tiden. I föreliggande avhandling skapas en holistisk översikt av kundernas karaktär och köpbeteende som utöver de konventionella segmenteringsmodellerna även beaktar dynamiken i köpbeteendet. Dynamiken i en kundsegmenteringsmodell innefattar förändringar i segmentens struktur och innehåll, samt förändringen av individuella kunders tillhörighet i ett segment (s.k migrationsanalyser). Vardera förändringen modelleras, analyseras och exemplifieras med visuella datautvinningstekniker, främst med självorganiserande kartor (SOM, eng. Self-Organizing Maps) och självorganiserande tidskartor (SOTM), en vidareutveckling av SOM. Visualiseringen anteciperas underlätta tolkningen av identifierade mönster och göra processen med kunskapsöverföring mellan den som gör analysen och beslutsfattaren smidigare. Asiakkuudenhallinta (CRM) eli organisaation muuttaminen tuotepainotteisesta asiakaskeskeiseksi on herättänyt mielenkiintoa niin yliopisto- kuin yritysmaailmassakin. Asiakkaiden käyttäytymistä ja toimintaa pystytään nykyään helposti tallentamaan ja varastoimaan toiminnanohjausjärjestelmien ja tietovarastojen avulla; asiakkaat jättävät jatkuvasti piirteistään ja ostokäyttäytymisestään kertovia tietojälkiä, joita voidaan analysoida. On tavallista, että asiakkaat poikkeavat toisistaan eri tavoin, ja heidän mieltymyksensä kuten myös ostokäyttäytymisensä saattavat olla hyvinkin erilaisia. Asiakaskäyttäytymisen monimuotoisuuteen ja tuottavuuteen paneuduttaessa käytetäänkin laajalti asiakassegmentointia eli asiakkaiden jakamista ryhmiin samankaltaisuuden perusteella. Perinteiset asiakassegmentoinnin ratkaisut ovat usein yksittäisiä analyyttisia malleja, jotka on tehty tietyn aikajakson perusteella. Tämän vuoksi ne monesti jättävät huomioimatta sen, että asiakkaiden käyttäytyminen saattaa ajan kuluessa muuttua. Tässä väitöskirjassa pyritäänkin tarjoamaan holistinen kuva asiakkaiden ominaisuuksista ja ostokäyttäytymisestä tarkastelemalla kahta muutosvoimaa tiettyyn aikarajaukseen perustuvien perinteisten segmentointimallien lisäksi. Nämä kaksi asiakassegmentointimallin dynamiikkaa ovat muutokset segmenttien rakenteessa ja muutokset yksittäisten asiakkaiden kuulumisessa ryhmään. Ensimmäistä dynamiikkaa lähestytään ajallisen asiakassegmentoinnin avulla, jossa visualisoidaan ajan kuluessa tapahtuvat muutokset segmenttien rakenteissa ja profiileissa. Toista dynamiikkaa taas lähestytään käyttäen nk. segmenttisiirtymien analyysia, jossa visuaalisin keinoin tunnistetaan samantyyppisesti segmentistä toiseen vaihtavat asiakkaat. Visualisoinnin tehtävänä on tukea havaittujen kaavojen tulkitsemista sekä helpottaa tiedonsiirtoa analysoijan ja päättäjien välillä. Visuaalisia tiedonlouhintamenetelmiä, kuten itseorganisoivia karttoja ja niiden laajennuksia, käytetään osoittamaan näiden menetelmien hyödyllisyys sekä asiakkuudenhallinnassa yleisesti että erityisesti asiakassegmentoinnissa.

Visual Arts Data Skills for Researchers (VADS4R)

Workshop at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

From KAPTUR to VADS4R: Exploring Research Data Management in the Visual Arts

Presentation at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

From KAPTUR to VADSR: Exploring Research Data Management in the Visual Arts

Poster at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

The neural processes generating visual phenomenal consciousness: ERP and neuronavigated brain stimulation studies

One of the greatest conundrums to the contemporary science is the relation between consciousness and brain activity, and one of the specifi c questions is how neural activity can generate vivid subjective experiences. Studies focusing on visual consciousness have become essential in solving the empirical questions of consciousness. Th e main aim of this thesis is to clarify the relation between visual consciousness and the neural and electrophysiological processes of the brain. By applying electroencephalography and functional magnetic resonance image-guided transcranial magnetic stimulation (TMS), we investigated the links between conscious perception and attention, the temporal evolution of visual consciousness during stimulus processing, the causal roles of primary visual cortex (V1), visual area 2 (V2) and lateral occipital cortex (LO) in the generation of visual consciousness and also the methodological issues concerning the accuracy of targeting TMS to V1. Th e results showed that the fi rst eff ects of visual consciousness on electrophysiological responses (about 140 ms aft er the stimulus-onset) appeared earlier than the eff ects of selective attention, and also in the unattended condition, suggesting that visual consciousness and selective attention are two independent phenomena which have distinct underlying neural mechanisms. In addition, while it is well known that V1 is necessary for visual awareness, the results of the present thesis suggest that also the abutting visual area V2 is a prerequisite for conscious perception. In our studies, the activation in V2 was necessary for the conscious perception of change in contrast for a shorter period of time than in the case of more detailed conscious perception. We also found that TMS in LO suppressed the conscious perception of object shape when TMS was delivered in two distinct time windows, the latter corresponding with the timing of the ERPs related to the conscious perception of coherent object shape. Th e result supports the view that LO is crucial in conscious perception of object coherency and is likely to be directly involved in the generation of visual consciousness. Furthermore, we found that visual sensations, or phosphenes, elicited by the TMS of V1 were brighter than identically induced phosphenes arising from V2. Th ese fi ndings demonstrate that V1 contributes more to the generation of the sensation of brightness than does V2. Th e results also suggest that top-down activation from V2 to V1 is probably associated with phosphene generation. The results of the methodological study imply that when a commonly used landmark (2 cm above the inion) is used in targeting TMS to V1, the TMS-induced electric fi eld is likely to be highest in dorsal V2. When V1 was targeted according to the individual retinotopic data, the electric fi eld was highest in V1 only in half of the participants. Th is result suggests that if the objective is to study the role of V1 with TMS methodology, at least functional maps of V1 and V2 should be applied with computational model of the TMS-induced electric fi eld in V1 and V2. Finally, the results of this thesis imply that diff erent features of attention contribute diff erently to visual consciousness, and thus, the theoretical model which is built up of the relationship between visual consciousness and attention should acknowledge these diff erences. Future studies should also explore the possibility that visual consciousness consists of several processing stages, each of which have their distinct underlying neural mechanisms.