86 research outputs found
Tracking complex primitives in an image sequence
No full textThis paper describes a new approach to track complex primitives along image sequences- integratingsnake-based contour tracking and region-based motion analysis. First, a snake tracks the region outline and performs segmentation. Then the motion of the extracted region is estimated by a dense analysis of the apparent motion over the region, using spatio-temporal image gradients. Finally, this motion measurement is filtered to predict the region location in the next frame, and thus to guide (i.e. to initialize) the tracking snake in the next frame. Therefore, these two approaches collaborate and exchange information to overcome the limitations of each of them. The method is illustrated by experimental results on real images.
Environmental stresses mediate endophyte–grass interactions in a boreal archipelago
No full textSummary
1. Both evolutionary theory and empirical evidence from agricultural research support the view that asexual, vertically transmitted fungal endophytes are typically plant mutualists that develop high infection frequencies within host grass populations. In contrast, endophyte–grass interactions in natural ecosystems are more variable, spanning the range from mutualism to antagonism and comparatively little is known about their range of response to environmental stress.
2. We examined patterns in endophyte prevalence and endophyte–grass interactions across nutrient and grazing (from Greylag and Canada geese) gradients in 15 sites with different soil moisture levels in 13 island populations of the widespread grass Festuca rubra in a boreal archipelago in Sweden.
3. In the field, endophyte prevalence levels were generally low (range = 10–53%) compared with those reported from agricultural systems. Under mesic‐moist conditions endophyte prevalence was constantly low (mean prevalence = 15%) and was not affected by grazing pressure or nutrient availability. In contrast, under conditions of drought, endophyte prevalence increased from 10% to 53% with increasing nutrient availability and increasing grazing pressure.
4. In the field, we measured the production of flowering culms, as a proxy for host fitness, to determine how endophyte‐infected plants differed from uninfected plants. At dry sites, endophyte infection did not affect flowering culm production. In contrast, at mesic‐moist sites production of flowering culms in endophyte‐infected plants increased with the covarying effects of increasing nutrient availability and grazing pressure, indicating that the interaction switched from antagonistic to mutualistic.
5. A concurrent glasshouse experiment showed that in most situations, the host appears to incur some costs for harbouring endophytes. Uninfected grasses generally outperformed infected grasses (antagonistic interaction), while infected grasses outperformed uninfected grasses (mutualistic interaction) only in dry, nutrient‐rich conditions. Nutrient and water addition affected tiller production, leaf number and leaf length differently, suggesting that tillers responded with different strategies. This emphasizes that several response variables are needed to evaluate the interaction.
6. Synthesis. This study found complex patterns in endophyte prevalence that were not always correlated with culm production. These contrasting patterns suggest that the direction and strength of selection on infected plants is highly variable and depends upon a suite of interacting environmental variables that may fluctuate in the intensity of their impact, during the course of the host life cycle.Peer reviewedfinal article publishedantagonist–mutualist continuumwater stresssymbiosisSwedennutrient stressherbivorygrassFestuca rubraEpichloë festucaeendophyt
One Step Forward and Two Steps Back: The Role of Civil Society Organizations in Reversed Integration Processes among Refugees in Norway
No full textThis is an open-access article distributed under the terms of the Creative Commons NonCommercial-NoDerivatives Attribution 4.0 International License (CC-BY-NC-ND 4.0), which permits unrestricted distribution, and reproduction in any medium, provided the original author and source are credited, the material is not used for commercial purposes and is not altered in any way. See https:// creativecommons.org/ licenses/by-nc-nd/4.0/Refugees settling in Norway experience several barriers to labour market integration, such as language insufficiency, low or unrecognised formal competence, and discrimination. While numerous scholars have dealt with the issues of both policy implementation and the outcome of public measures, there is a growing interest in the contributions of civil society organizations (CSOs) to the labour market integration of refugees. Such an interest is fueled by the proliferation of neoliberal reforms in European welfare states and restrictive budgets, leading to increased recognition of CSOs as resolutions to social issues. Based on ethnographic fieldwork among refugees and employees in two CSOs in Norway, the findings suggest a particularly vulnerable phase immediately after the public introduction programme for refugees not moving on to employment, education, or training. Drawing on Bourdieu’s concept of different forms of capital in addition to Granovetter’s theory of social network, I argue that CSOs have a profound role in preventing the reversed integration processes that occur in this specific phase of settlement.publishedVersio
Deep Multi-View Stereo Gone Wild
No full textInternational audienceDeep multi-view stereo (MVS) methods have been developed and extensively compared on simple datasets, where they now outperform classical approaches. In this paper, we ask whether the conclusions reached in controlled scenarios are still valid when working with Internet photo collections. We propose a methodology for evaluation and explore the influence of three aspects of deep MVS methods: network architecture, training data, and supervision. We make several key observations, which we extensively validate quantitatively and qualitatively, both for depth prediction and complete 3D reconstructions. First, complex unsupervised approaches cannot train on data in the wild. Our new approach makes it possible with three key elements: upsampling the output, softmin based aggregation and a single reconstruction loss. Second, supervised deep depthmap-based MVS methods are state-of-the art for reconstruction of few internet images. Finally, our evaluation provides very different results than usual ones. This shows that evaluation in uncontrolled scenarios is important for new architectures
- …
