OuluREPO University of Oulu repository
Not a member yet
54977 research outputs found
Sort by
Data-driven severity prediction of net blotch in spring barley
No full textAbstract
The use of computational methods has increased in recent years in the field of plant disease research and prevention. This study focused on the data-driven prediction of barley net blotch severity in spring barley using weather data. By analysing 36 data sets spanning 26 years of publicly available weather measurements together with historical barley net blotch severity observations, the information content of the weather data was enriched by constructing computational features from the measurements using various mathematical operations. The predictability of the barley net blotch severity was investigated using the derived features and different mathematical classifiers.
The findings strongly suggest that weather data can be used to accurately predict net blotch occurrence during the early growing season in Finland. Moreover, the information content of meteorological data varies throughout the season. Notably, the optimal starting point for data-driven prediction can be set automatically by identifying the onset of the growing season based on the outdoor temperature.
The methodology developed employs computational techniques with feature engineering, removing the need for additional measurement campaigns. Ensemble classifiers, composed of binary linear discriminant models and aggregated using the geometric mean, consistently outperformed other approaches – achieving near-perfect accuracy. This approach supports optimised pesticide application, reducing both environmental impact and costs, and aligns with sustainable agricultural practices. Original papers Ruusunen, O., Jalli, M., Jauhiainen, L., Ruusunen, M., & Leiviskä, K. (2020). Advanced data analysis as a tool for net blotch density estimation in spring barley. Agriculture, 10(5), 179. https://doi.org/10.3390/agriculture10050179 https://doi.org/10.3390/agriculture10050179 Self-archived version Ruusunen, O., Jalli, M., Jauhiainen, L., Ruusunen, M., & Leiviskä, K. (2022). Identification of optimal starting time instance to forecast net blotch density in spring barley with meteorological data in Finland. Agriculture, 12(11), 1939. https://doi.org/10.3390/agriculture12111939 https://doi.org/10.3390/agriculture12111939 Self-archived version Ruusunen, O., Jalli, M., Jauhiainen, L., Ruusunen, M., & Leiviskä, K. (2024). Linear discriminant analysis for predicting net blotch severity in spring barley with meteorological data in Finland. Agriculture, 14(10), 1779. https://doi.org/10.3390/agriculture14101779 https://doi.org/10.3390/agriculture14101779 Self-archived version Tiivistelmä
Laskennallisten menetelmien soveltaminen on yleistynyt kasvitautien tutkimuksessa sekä torjunnassa. Tässä tutkimuksessa keskityttiin kevätohran verkkolaikun tautiasteen datapohjaiseen ennustamiseen säätietojen avulla. Olemassa olevien säähavaintojen informaatiosisältöä hyödynnettiin muodostamalla mittausaineistosta laskennallisia piirteitä erilaisten matemaattisten operaatioiden avulla. Verkkolaikun tautiasteen ennustettavuutta tutkittiin muodostettujen piirteiden sekä erilaisten matemaattisten luokittelijoiden avulla. Työssä analysoitiin julkisia säätietoja sekä vastaavia havaintoaineistoja ohran verkkolaikun tautiasteesta 26 vuoden ajalta. Käytettävissä oli yhteensä 36 erillistä säämuuttujista ja ohran verkkolaikun tautiasteesta muodostettua mittausaineistoa.
Tulokset osoittavat, että meteorologiset mittaukset sisältävät riittävästi tietoa kevätohran verkkolaikun esiintymisen tehokkaaseen ennustamiseen kasvukauden alkuvaiheessa Suomessa. Mittausaineiston informaatiosisältö kuitenkin vaihtelee kasvukauden aikana. Lisäksi havaittiin, että ennusteen aloitusajankohta voidaan määrittää automaattisesti kasvukauden alkuun ulkolämpötilamittauksiin perustuen.
Kehitetty ennustemenetelmä perustuu matemaattiseen luokitteluun ja säätiedoista muodostettujen piirteiden käyttöön, mikä poistaa erillisten mittauskampanjoiden tarpeen. Lähes täydellinen ennustetarkkuus saavutettiin yhdistelmämalleilla, joissa käytettiin binäärisiä lineaarisia diskriminanttiluokittelijoita ja niiden tulosten yhdistämistä geometrisen keskiarvon avulla. Luotettava ennustemenetelmä mahdollistaa torjunta-aineiden käytön optimoinnin, mikä vähentää ympäristökuormitusta ja kustannuksia sekä edistää kestävää maataloutta. Osajulkaisut Ruusunen, O., Jalli, M., Jauhiainen, L., Ruusunen, M., & Leiviskä, K. (2020). Advanced data analysis as a tool for net blotch density estimation in spring barley. Agriculture, 10(5), 179. https://doi.org/10.3390/agriculture10050179 https://doi.org/10.3390/agriculture10050179 Rinnakkaistallennettu versio Ruusunen, O., Jalli, M., Jauhiainen, L., Ruusunen, M., & Leiviskä, K. (2022). Identification of optimal starting time instance to forecast net blotch density in spring barley with meteorological data in Finland. Agriculture, 12(11), 1939. https://doi.org/10.3390/agriculture12111939 https://doi.org/10.3390/agriculture12111939 Rinnakkaistallennettu versio Ruusunen, O., Jalli, M., Jauhiainen, L., Ruusunen, M., & Leiviskä, K. (2024). Linear discriminant analysis for predicting net blotch severity in spring barley with meteorological data in Finland. Agriculture, 14(10), 1779. https://doi.org/10.3390/agriculture14101779 https://doi.org/10.3390/agriculture14101779 Rinnakkaistallennettu versio Academic dissertation to be presented with the assent of the Doctoral Programme Committee of Technology and Natural Sciences of the University of Oulu for public defence in the OP-Pohjola auditorium (L6), Linnanmaa, on 6 February 2026, at 12 noonAbstract
The use of computational methods has increased in recent years in the field of plant disease research and prevention. This study focused on the data-driven prediction of barley net blotch severity in spring barley using weather data. By analysing 36 data sets spanning 26 years of publicly available weather measurements together with historical barley net blotch severity observations, the information content of the weather data was enriched by constructing computational features from the measurements using various mathematical operations. The predictability of the barley net blotch severity was investigated using the derived features and different mathematical classifiers.
The findings strongly suggest that weather data can be used to accurately predict net blotch occurrence during the early growing season in Finland. Moreover, the information content of meteorological data varies throughout the season. Notably, the optimal starting point for data-driven prediction can be set automatically by identifying the onset of the growing season based on the outdoor temperature.
The methodology developed employs computational techniques with feature engineering, removing the need for additional measurement campaigns. Ensemble classifiers, composed of binary linear discriminant models and aggregated using the geometric mean, consistently outperformed other approaches – achieving near-perfect accuracy. This approach supports optimised pesticide application, reducing both environmental impact and costs, and aligns with sustainable agricultural practices.Tiivistelmä
Laskennallisten menetelmien soveltaminen on yleistynyt kasvitautien tutkimuksessa sekä torjunnassa. Tässä tutkimuksessa keskityttiin kevätohran verkkolaikun tautiasteen datapohjaiseen ennustamiseen säätietojen avulla. Olemassa olevien säähavaintojen informaatiosisältöä hyödynnettiin muodostamalla mittausaineistosta laskennallisia piirteitä erilaisten matemaattisten operaatioiden avulla. Verkkolaikun tautiasteen ennustettavuutta tutkittiin muodostettujen piirteiden sekä erilaisten matemaattisten luokittelijoiden avulla. Työssä analysoitiin julkisia säätietoja sekä vastaavia havaintoaineistoja ohran verkkolaikun tautiasteesta 26 vuoden ajalta. Käytettävissä oli yhteensä 36 erillistä säämuuttujista ja ohran verkkolaikun tautiasteesta muodostettua mittausaineistoa.
Tulokset osoittavat, että meteorologiset mittaukset sisältävät riittävästi tietoa kevätohran verkkolaikun esiintymisen tehokkaaseen ennustamiseen kasvukauden alkuvaiheessa Suomessa. Mittausaineiston informaatiosisältö kuitenkin vaihtelee kasvukauden aikana. Lisäksi havaittiin, että ennusteen aloitusajankohta voidaan määrittää automaattisesti kasvukauden alkuun ulkolämpötilamittauksiin perustuen.
Kehitetty ennustemenetelmä perustuu matemaattiseen luokitteluun ja säätiedoista muodostettujen piirteiden käyttöön, mikä poistaa erillisten mittauskampanjoiden tarpeen. Lähes täydellinen ennustetarkkuus saavutettiin yhdistelmämalleilla, joissa käytettiin binäärisiä lineaarisia diskriminanttiluokittelijoita ja niiden tulosten yhdistämistä geometrisen keskiarvon avulla. Luotettava ennustemenetelmä mahdollistaa torjunta-aineiden käytön optimoinnin, mikä vähentää ympäristökuormitusta ja kustannuksia sekä edistää kestävää maataloutta
Satelliitti- ja maanpäällisten 5G RedCap -verkkojen vertaileva elinkaarianalyysi
No full textFifth Generation (5G) telecommunication networks are being rapidly deployed worldwide, raising concerns about their environmental implications. Operating on high-frequency bands, 5G requires denser infrastructure, leading to increased energy consumption, material use, and associated greenhouse gas emissions. In parallel, modern connectivity is evolving through the integration of Non-Terrestrial Networks (NTNs) employing Low Earth Orbit satellites. Both 5G Terrestrial Network (TN) and NTN architectures support Reduced Capability (RedCap) devices, which enable large-scale industrial IoT connectivity with reduced energy consumption and device complexity.
As connectivity infrastructure expands, understanding its environmental impacts is essential for sustainable network design. This thesis investigates the environmental trade-offs between 5G TN and NTN systems using a Life Cycle Assessment (LCA) approach. The assessment focuses on radio access network nodes, with the functional unit defined as providing equivalent 5G RedCap connectivity for one year. The systems are evaluated across manufacturing, deployment (or launch), and operation stages, considering four impact categories: Climate Change, Resource Depletion, Non-Renewable Energy Demand, and Freshwater Eutrophication Potential. LCI modelling is conducted using SimaPro and OpenLCA, drawing on space-system datasets from the European Space Agency and the Strathclyde Space Systems Database, as well as site-specific datasets for TN.
The results show that TN exhibits the highest impacts in Non-Renewable Energy Demand and Freshwater Eutrophication Potential, driven mainly by continuous operational energy demand. In contrast, NTN shows higher impacts in Climate Change and Resource Depletion due to launch-related emissions and the use of rare materials in satellite solar panel assembly. By clarifying the relative environmental performance of terrestrial and satellite-based networks, this work supports more informed decision-making in telecommunication infrastructure planning and identifies priorities for future research, including improved data availability and expanded LCA scope
Non-invasive chemical characterisation of archaeological ochres from the early 4th millennium BCE forager graves and settlements in Finland
No full textAbstract
In the early 4th millennium BCE, the region of modern-day Finland was inhabited by pottery-using foragers, the Typical Comb Ware culture (TCW) people, who lived in village-like clusters of semi-subterranean houses and buried their dead in ochre-coloured graves. Here, we report the results of the analyses of archaeological ochre samples from burial and domestic contexts at eight TCW sites across Finland using non-invasive energy-dispersive X-ray fluorescence (ED-XRF) and scanning electron microscopy with energy dispersive spectrometry (SEM-EDS). Our comparative characterisation of the archaeological ochres aimed to distinguish whether 1) a signature ochre composition per site/community could be chemically discriminated; 2) similar ochre chemical fingerprints are detected in both burial and domestic settings in specific regional contexts; 3) the chemical data indicates shared ochre procurement strategies or ochre exchange among the studied communities or regions; and 4) how ochre was used as part of TCW mortuary practices. Based on our study, non-invasive ED-XRF of archaeological ochre samples enables the characterisation of a limited set of elemental concentrations; however, statistical analysis of log10 Fe-normalised concentration values allowed us to identify compositional groups in our dataset that correlate with the site location clusters or specific archaeological or geological phenomena. Hence, we propose that this result indicates regionalised ochre procurement and, in some cases, inter-regional, two-way transport of ochre, which may have specific, cross-regionally desired characteristics. Simultaneously, our results also highlight the deliberate use of different types of ochre within ritualized practices.Abstract
In the early 4th millennium BCE, the region of modern-day Finland was inhabited by pottery-using foragers, the Typical Comb Ware culture (TCW) people, who lived in village-like clusters of semi-subterranean houses and buried their dead in ochre-coloured graves. Here, we report the results of the analyses of archaeological ochre samples from burial and domestic contexts at eight TCW sites across Finland using non-invasive energy-dispersive X-ray fluorescence (ED-XRF) and scanning electron microscopy with energy dispersive spectrometry (SEM-EDS). Our comparative characterisation of the archaeological ochres aimed to distinguish whether 1) a signature ochre composition per site/community could be chemically discriminated; 2) similar ochre chemical fingerprints are detected in both burial and domestic settings in specific regional contexts; 3) the chemical data indicates shared ochre procurement strategies or ochre exchange among the studied communities or regions; and 4) how ochre was used as part of TCW mortuary practices. Based on our study, non-invasive ED-XRF of archaeological ochre samples enables the characterisation of a limited set of elemental concentrations; however, statistical analysis of log10 Fe-normalised concentration values allowed us to identify compositional groups in our dataset that correlate with the site location clusters or specific archaeological or geological phenomena. Hence, we propose that this result indicates regionalised ochre procurement and, in some cases, inter-regional, two-way transport of ochre, which may have specific, cross-regionally desired characteristics. Simultaneously, our results also highlight the deliberate use of different types of ochre within ritualized practices
Relation between porosity and reduction degree of iron ore pellets in hydrogen-assisted reduction
No full textAttitudes of the care staff towards patients in Finnish mental hospitals from 1970 to 1975
No full textPro Gradu -tutkielmassani tutkin mielisairaaloiden ja niissä hoidettujen potilaiden historiaa 1970-luvun ensimmäisellä puoliskolla. Kiinnostukseni kohdistuu ennen kaikkea hoitohenkilökunnan asenteisiin mielisairaaloissa hoidettuja potilaita kohtaan. Tutkimuskysymykseni on: Miten hoitohenkilökunta Oulun keskusmielisairaalassa ja psykiatrit Duodecim-lehdessä ja psykiatrian oppikirjassa asennoituivat mielisairaalan potilaisiin vuosina 1970—1975? Lähteenä toimivat niin ikään Oulun keskusmielisairaalan tiedotuslehti ja Duodecim –lehden psykiatriaa koskevat artikkelit vuosilta 1970—1975 sekä vuoden 1971 psykiatrian oppikirja (Psykiatria).
Menetelmänä tutkielmassa käytän historiallista diskurssianalyysia. Tutkin tekstin ja kielen tuottamia merkityksiä ja ennen kaikkea diskursseissa esiintyviä valtasuhteita. Tärkein käsite tutkielmassa on asenne, jota tulkitsen Martin Fishbeinin ja Icek Ajzenin sosiaalipsykologisen asenne –teorian viitekehyksessä. Lisäksi läpi tutkielman kulkevina teemoina ovat Erving Goffmanin 1960-luvulla kehittelemät totaalisen laitoksen ja stigman käsitteet.
Tutkielma etenee temaattisesti ja jakautuu kahteen lukuun. Ensimmäisessä luvussa teemana on potilas ja hoitajat laitoksissa. Toinen luku puolestaan suuntaa katseensa sairaalan ja laitoshoidon jälkeiseen aikaan: potilaan paluuseen yhteiskuntaan. Toisen luvun teemaan liittyy aiheina vahvasti kuntoutus ja avohoito, jotka olivat 1970-luvun psykiatrisen hoidon isoja käännekohtia. Pohdin läpi tutkielman, miten potilaiden toimijuus näyttäytyy eri teemojen alla.
Tutkimuskysymyksiini ei ole mahdollista löytää yhtä selkeää vastausta lähteiden ja tutkimuskirjallisuuden pohjalta, sillä nähdäkseni tutkimanani aikana ei ollut varsinaista konsensusta siitä, miten potilaisiin tulisi asennoitua. Minulle jää sellainen käsitys, että erilaisia asenteita mielisairaita kohtaan oli melkein yhtä paljon kuin kirjoittajia. Kuitenkin jonkinlaisia suuria linjoja on mahdollista löytää. Esimerkiksi kirjoittajan (auktoriteetti)asema nähtävästi vaikutti tämän suhtautumiseen. Lääkärit kirjoittivat useammin ylhäältä alaspäin, jopa alentavaan sävyyn, kun taas hoitajat näkivät potilaat jokapäiväisessä arjessa, minkä vuoksi suhtautuminen saattoi olla lempeämpää. Myös sillä oli vaikutusta, minkä kaltaisessa julkaisussa potilaista kirjoitettiin: tiedotuslehdistä välittyi usein lämpimämpi suhtautuminen verrattuna oppikirjaan ja Duodecim-lehtiin, mitkä olivat suunnattu ammattilaisten luettavaksi ja asenteet jopa verhoutuivat toisinaan objektiivisuuden taakse piiloon
Enhancement of Surface Integrity of Binder Jet Fabricated Stainless Steel 316L via Severe Shot Peening
Get PDFAbstract
This study investigates the effects of heat treatment (HT) and severe shot peening (SSP) on the surface integrity of binder jetting (BJ) manufactured 316L stainless steel. While HT step was chosen for its proven effectiveness in relieving residual stresses in PBF-LB built 316L, it was observed to increase porosity in BJ samples from 2.5% to 7.5%. SSP alone, however, effectively enhanced surface hardness from 145 to 504 HV, introduced beneficial compressive residual stresses reaching −995 MPa at a depth of 91 μm (remaining compressive up to 300 μm), and reduced surface porosity to 0.45%. These improvements indicate a significant enhancement in surface integrity, thus potentially improving wear and fatigue resistance. The findings suggest that SSP is sufficient for optimizing surface properties in BJ components, offering an effective post-processing approach for high-performance applications.Abstract
This study investigates the effects of heat treatment (HT) and severe shot peening (SSP) on the surface integrity of binder jetting (BJ) manufactured 316L stainless steel. While HT step was chosen for its proven effectiveness in relieving residual stresses in PBF-LB built 316L, it was observed to increase porosity in BJ samples from 2.5% to 7.5%. SSP alone, however, effectively enhanced surface hardness from 145 to 504 HV, introduced beneficial compressive residual stresses reaching −995 MPa at a depth of 91 μm (remaining compressive up to 300 μm), and reduced surface porosity to 0.45%. These improvements indicate a significant enhancement in surface integrity, thus potentially improving wear and fatigue resistance. The findings suggest that SSP is sufficient for optimizing surface properties in BJ components, offering an effective post-processing approach for high-performance applications
Kaleidoscopic AI: Hallucinations on the Verge of Creativity
No full textAbstract
Artificial Intelligence (AI) has garnered significant attention, especially in regard to imagination, particularly due to recent advancements in text and image generation. However, it has also raised numerous concerns regarding the dissemination of misinformation, the phenomenon of hallucinations, and the generation of fantastical narratives that diverge from actual reality. This analysis will examine research on generative AI hallucinations and will seek to expand upon their findings by exploring the potential for design imagination arising from these phenomena, while also proposing a philosophical inquiry into the reasons behind AI’s generation of hallucinations in response to our creative requests and comparing it to child development to reflect on the evolution of human imagination. This study will establish the groundwork for an alternative perspective on generative AI and artificial intelligence overall, serving as a reflection of our endeavors that presents the world in an innovative manner, thereby offering a fresh and radiant lens for crafting creative narratives to influence it.Abstract
Artificial Intelligence (AI) has garnered significant attention, especially in regard to imagination, particularly due to recent advancements in text and image generation. However, it has also raised numerous concerns regarding the dissemination of misinformation, the phenomenon of hallucinations, and the generation of fantastical narratives that diverge from actual reality. This analysis will examine research on generative AI hallucinations and will seek to expand upon their findings by exploring the potential for design imagination arising from these phenomena, while also proposing a philosophical inquiry into the reasons behind AI’s generation of hallucinations in response to our creative requests and comparing it to child development to reflect on the evolution of human imagination. This study will establish the groundwork for an alternative perspective on generative AI and artificial intelligence overall, serving as a reflection of our endeavors that presents the world in an innovative manner, thereby offering a fresh and radiant lens for crafting creative narratives to influence it
Fixed-time convergence ZNN model for solving rectangular dynamic full-rank matrices inversion
No full textAbstract
The Moore–Penrose inverse of dynamic matrices has found widespread application and has garnered significant attention. The zeroing neural network (ZNN) has proven to be an effective solution for computing the Moore–Penrose inverse in dynamic matrices. This paper proposes a novel unified fixed-time ZNN (UFTZNN) model designed to achieve fixed-time convergence and solve both left and right inverse problems using a single model. Theoretical analysis of the convergence and robustness of the UFTZNN model is rigorously presented. Numerical simulations comparing the UFTZNN with existing ZNN models confirm its superiority in addressing left and right inverse problems, convergence time, and robustness. The UFTZNN model is applied to the inverse kinematic tracking problem of a six-degree-of-freedom manipulator-based photoelectric tracking system to demonstrate its potential applications and effectiveness.Abstract
The Moore–Penrose inverse of dynamic matrices has found widespread application and has garnered significant attention. The zeroing neural network (ZNN) has proven to be an effective solution for computing the Moore–Penrose inverse in dynamic matrices. This paper proposes a novel unified fixed-time ZNN (UFTZNN) model designed to achieve fixed-time convergence and solve both left and right inverse problems using a single model. Theoretical analysis of the convergence and robustness of the UFTZNN model is rigorously presented. Numerical simulations comparing the UFTZNN with existing ZNN models confirm its superiority in addressing left and right inverse problems, convergence time, and robustness. The UFTZNN model is applied to the inverse kinematic tracking problem of a six-degree-of-freedom manipulator-based photoelectric tracking system to demonstrate its potential applications and effectiveness
Testability-Driven Development: An Improvement to the TDD Efficiency
No full textAbstract
Test-first development (TFD) is a software development approach involving automated tests before writing the actual code. TFD offers many benefits, such as improving code quality, reducing debugging time, and enabling easier refactoring. However, TFD also poses challenges and limitations, requiring more effort and time to write and maintain test cases, especially for large and complex projects. Refactoring for testability is improving the internal structure of source code to make it easier to test. Refactoring for testability can reduce the cost and complexity of software testing and speed up the test-first life cycle. However, measuring testability is a vital step before refactoring for testability, as it provides a baseline for evaluating the current state of the software and identifying the areas that need improvement. This paper proposes a mathematical model for calculating class testability based on test effectiveness and effort and a machine-learning regression model that predicts testability using source code metrics. It also introduces a testability-driven development (TsDD) method that conducts the TFD process toward developing testable code. TsDD focuses on improving testability and reducing testing costs by measuring testability frequently and refactoring to increase testability without running the program. Our testability prediction model has a mean squared error of 0.0311 and an R2 score of 0.6285. We illustrate the usefulness of TsDD by applying it to 50 Java classes from three open-source projects. TsDD achieves an average of 77.81% improvement in the testability of these classes. Experts’ manual evaluation confirms the potential of TsDD in accelerating the TDD process.Abstract
Test-first development (TFD) is a software development approach involving automated tests before writing the actual code. TFD offers many benefits, such as improving code quality, reducing debugging time, and enabling easier refactoring. However, TFD also poses challenges and limitations, requiring more effort and time to write and maintain test cases, especially for large and complex projects. Refactoring for testability is improving the internal structure of source code to make it easier to test. Refactoring for testability can reduce the cost and complexity of software testing and speed up the test-first life cycle. However, measuring testability is a vital step before refactoring for testability, as it provides a baseline for evaluating the current state of the software and identifying the areas that need improvement. This paper proposes a mathematical model for calculating class testability based on test effectiveness and effort and a machine-learning regression model that predicts testability using source code metrics. It also introduces a testability-driven development (TsDD) method that conducts the TFD process toward developing testable code. TsDD focuses on improving testability and reducing testing costs by measuring testability frequently and refactoring to increase testability without running the program. Our testability prediction model has a mean squared error of 0.0311 and an R2 score of 0.6285. We illustrate the usefulness of TsDD by applying it to 50 Java classes from three open-source projects. TsDD achieves an average of 77.81% improvement in the testability of these classes. Experts’ manual evaluation confirms the potential of TsDD in accelerating the TDD process
Discovering attention-guided cross-modality correlation for visible–infrared person re-identification
No full textAbstract
Visible–infrared person re-identification (VI Re-ID) is an essential and challenging task. Existing studies mainly focus on learning the unified modality-invariant representations directly from visible and infrared images. However, it is hard to obtain the identity-aware patterns due to the co-existence of inter- and intra-modality discrepancies. In this paper, we propose a novel attention-guided cross-modality correlation method (AGCC) to achieve the modality-invariant and identity-discriminative representations for visible–infrared person Re-ID. Specifically, we introduce a modality-aware attention (MAA) mechanism to model the inter- and intra-modality variations, which generates attention masks of two modalities for preserving the most significant region and obtaining the discriminative patterns in each identity. Further, we present an attention-guided channel and spatial correlation scheme (AGCSC) to establish the attention-guided cross-modality correlation, which can bridge the gap between inter- and intra-modalities. Moreover, a novel joint-modality learning head (JMLH) is developed to promote the metric and mutual learning from both feature distribution and classification logit levels. Extensive experiments on two public SYSU-MM01 and RegDB datasets demonstrate the remarkable superiority of our method over the state of the arts. The implementation codes will be made available soon.Abstract
Visible–infrared person re-identification (VI Re-ID) is an essential and challenging task. Existing studies mainly focus on learning the unified modality-invariant representations directly from visible and infrared images. However, it is hard to obtain the identity-aware patterns due to the co-existence of inter- and intra-modality discrepancies. In this paper, we propose a novel attention-guided cross-modality correlation method (AGCC) to achieve the modality-invariant and identity-discriminative representations for visible–infrared person Re-ID. Specifically, we introduce a modality-aware attention (MAA) mechanism to model the inter- and intra-modality variations, which generates attention masks of two modalities for preserving the most significant region and obtaining the discriminative patterns in each identity. Further, we present an attention-guided channel and spatial correlation scheme (AGCSC) to establish the attention-guided cross-modality correlation, which can bridge the gap between inter- and intra-modalities. Moreover, a novel joint-modality learning head (JMLH) is developed to promote the metric and mutual learning from both feature distribution and classification logit levels. Extensive experiments on two public SYSU-MM01 and RegDB datasets demonstrate the remarkable superiority of our method over the state of the arts. The implementation codes will be made available soon