Search

International audience; We address the issue of speeding up the training of convolutional neural networks by studying a distributed method adapted to stochastic gradient descent. Our parallel optimization setup uses several threads, each applying individual gradient descents on a local variable. We propose a new way of sharing information between different threads based on gossip algorithms that show good consensus convergence properties. Our method called GoSGD has the advantage to be fully asynchronous and decentralized.

In this paper, we show how uncertainty estimation can be leveraged to enable safety critical image segmentation in autonomous driving, by triggering a fallback behavior if a target accuracy cannot be guaranteed. We introduce a new uncertainty measure based on disagreeing predictions as measured by a dissimilarity function. We propose to estimate this dissimilarity by training a deep neural architecture in parallel to the task-specific network. It allows this observer to be dedicated to the uncertainty estimation, and let the task-specific network make predictions. We propose to use self-supervision to train the observer, which implies that our method does not require additional training data. We show experimentally that our proposed approach is much less computationally intensive at inference time than competing methods (e.g. MCDropout), while delivering better results on safety-oriented evaluation metrics on the CamVid dataset, especially in the case of glare artifacts.

Solar Orbiter is a solar mission that will approach the Sun down to a minimum perihelion of 0.28 AU and will increase its orbit inclination with respect to the ecliptic up to a maximum angle of 34 deg. For imagers aboard Solar Orbiter there will be three 10-days remote sensing windows per orbit. Observations shall be carefully planned at least 6 months in advance. The Multi Instrument Sequence Organizer (MISO) is a web based platform developed by the SPICE group and made available to support Solar Orbiter instruments teams in planning observations by assembling Mission Database sequences. Metis is the UV and visible light coronagraph aboard Solar Orbiter. Metis is a complex instrument characterized by a rich variety of observing modes, which required a careful commissioning activity and will need support for potential maintenance operations throughout the mission. In order to support commissioning and maintenance activities, the Metis team developed a PDOR (Payload Direct Operation Request) and MDOR (Memory Direct Operation Request) module integrated in MISO and made available to all Solar Orbiter instruments. An effort was made in order to interpret the coding philosophy of the main project and to make the additional module as homogeneous as possible both to the web interface and to the algorithm logic, while integrating characteristics which are peculiar to PDORs and MDORs. An user friendly web based interface allows the operator to build the operation request and to successively modify or integrate it with further or alternative information. In the present work we describe the PDOR/MDOR module for MISO by addressing its logic and main characteristics.

Recent breakthroughs in representation learning of unseen classes and examples have been made in deep metric learning by training at the same time the image representations and a corresponding metric with deep networks. Recent contributions mostly address the training part (loss functions, sampling strategies, etc.), while a few works focus on improving the discriminative power of the image representation. In this paper, we propose DIABLO, a dictionary-based attention method for image embedding. DIABLO produces richer representations by aggregating only visually-related features together while being easier to train than other attention-based methods in deep metric learning. This is experimentally confirmed on four deep metric learning datasets (Cub-200-2011, Cars-196, Stanford Online Products, and In-Shop Clothes Retrieval) for which DIABLO shows state-of-the-art performances., Pre-print. Accepted for publication at Pattern Recognition Letters

We tackle the image reassembly problem with wide space between the fragments, in such a way that the patterns and colors continuity is mostly unusable. The spacing emulates the erosion of which the archaeological fragments suffer. We crop-square the fragments borders to compel our algorithm to learn from the content of the fragments. We also complicate the image reassembly by removing fragments and adding pieces from other sources. We use a two-step method to obtain the reassemblies: 1) a neural network predicts the positions of the fragments despite the gaps between them; 2) a graph that leads to the best reassemblies is made from these predictions. In this paper, we notably investigate the effect of branch-cut in the graph of reassemblies. We also provide a comparison with the literature, solve complex images reassemblies, explore at length the dataset, and propose a new metric that suits its specificities. Keywords: image reassembly, jigsaw puzzle, deep learning, graph, branch-cut, cultural heritage

International audience; In this paper, we tackle the problem of human pose estimation from still images, which is a very active topic, specially due to its several applications, from image annotation to human-machine interface. We use the soft-argmax function to convert feature maps directly to body joint coordinates, resulting in a fully differentiable framework. Our method is able to learn heat maps representations indirectly, without additional steps of artificial ground truth generation. Consequently, contextual information can be included to the pose predictions in a seamless way. We evaluated our method on two challenging datasets, the Leeds Sports Poses (LSP) and the MPII Human Pose datasets, reaching the best performance among all the existing regression methods. Source code available at: https://github.com/dluvizon/pose-regression.

Learning an effective similarity measure between image representations is key to the success of recent advances in visual search tasks (e.g. verification or zero-shot learning). Although the metric learning part is well addressed, this metric is usually computed over the average of the extracted deep features. This representation is then trained to be discriminative. However, these deep features tend to be scattered across the feature space. Consequently, the representations are not robust to outliers, object occlusions, background variations, etc. In this paper, we tackle this scattering problem with a distribution-aware regularization named HORDE. This regularizer enforces visually-close images to have deep features with the same distribution which are well localized in the feature space. We provide a theoretical analysis supporting this regularization effect. We also show the effectiveness of our approach by obtaining state-of-the-art results on 4 well-known datasets (Cub-200-2011, Cars-196, Stanford Online Products and Inshop Clothes Retrieval)., Camera-ready for our ICCV 2019 paper (poster)

Learning rich and compact representations is an open topic in many fields such as object recognition or image retrieval. Deep neural networks have made a major breakthrough during the last few years for these tasks but their representations are not necessary as rich as needed nor as compact as expected. To build richer representations, high order statistics have been exploited and have shown excellent performances, but they produce higher dimensional features. While this drawback has been partially addressed with factorization schemes, the original compactness of first order models has never been retrieved, or at the cost of a strong performance decrease. Our method, by jointly integrating codebook strategy to factorization scheme, is able to produce compact representations while keeping the second order performances with few additional parameters. This formulation leads to state-of-the-art results on three image retrieval datasets., Comment: Accepted at IEEE International Conference on Image Processing (ICIP) 2019

This article explores strangeness-familiarity relations in the context of global wine consumer culture. While observing wine as a deeply familiar foodstuff and consumer commodity that is part of th...

Action recognition and human pose estimation are closely related but both problems are generally handled as distinct tasks in the literature. In this work, we propose a multitask framework for jointly 2D and 3D pose estimation from still images and human action recognition from video sequences. We show that a single architecture can be used to solve the two problems in an efficient way and still achieves state-of-the-art results. Additionally, we demonstrate that optimization from end-to-end leads to significantly higher accuracy than separated learning. The proposed architecture can be trained with data from different categories simultaneously in a seamlessly way. The reported results on four datasets (MPII, Human3.6M, Penn Action and NTU) demonstrate the effectiveness of our method on the targeted tasks., To appear in CVPR 2018

Drawing on ethnographic data collected among a group of coastal villagers in Southwest Madagascar between 2006 and 2011, this article looks at the relationship between hospitality towards strangers and land dispossession, a major issue of concern in Madagascar, especially given the comparatively recent mineral, energy resource, and stone mining rush in the country. Belief in spirits plays a role in this.

Most image retrieval methods use global features that aggregate local distinctive patterns into a single representation. However, the aggregation process destroys the relative spatial information by considering orderless sets of local descriptors. We propose to integrate relative spatial information into the aggregation process by taking into account co-occurrences of local patterns in a tensor framework. The resulting signature called Improved Spatial Tensor Aggregation (ISTA) is able to reach state of the art performances on well known datasets such as Holidays, Oxford5k and Paris6k., Comment: 8 pages, 2 figures and 1 table. Draft paper for conference, IEEE International Conference on Image Processing (ICIP) 2018

This paper addresses the problem of reassembling images from disjointed fragments. More specifically, given an unordered set of fragments, we aim at reassembling one or several possibly incomplete images. The main contributions of this work are: 1) several deep neural architectures to predict the relative position of image fragments that outperform the previous state of the art; 2) casting the reassembly problem into the shortest path in a graph problem for which we provide several construction algorithms depending on available information; 3) a new dataset of images taken from the Metropolitan Museum of Art (MET) dedicated to image reassembly for which we provide a clear setup and a strong baseline., ECCV 2018

Designing powerful tools that support cooking activities has rapidly gained popularity due to the massive amounts of available data, as well as recent advances in machine learning that are capable of analyzing them. In this paper, we propose a cross-modal retrieval model aligning visual and textual data (like pictures of dishes and their recipes) in a shared representation space. We describe an effective learning scheme, capable of tackling large-scale problems, and validate it on the Recipe1M dataset containing nearly 1 million picture-recipe pairs. We show the effectiveness of our approach regarding previous state-of-the-art models and present qualitative results over computational cooking use cases., Comment: accepted at the 41st International ACM SIGIR Conference on Research and Development in Information Retrieval, 2018

Archaeologists are in dire need of automated object reconstruction methods. Fragments reassembly is close to puzzle problems, which may be solved by computer vision algorithms. As they are often beaten on most image related tasks by deep learning algorithms, we study a classification method that can solve jigsaw puzzles. In this paper, we focus on classifying the relative position: given a couple of fragments, we compute their local relation (e.g. on top). We propose several enhancements over the state of the art in this domain, which is outperformed by our method by 25\%. We propose an original dataset composed of pictures from the Metropolitan Museum of Art. We propose a greedy reconstruction method based on the predicted relative positions., Comment: ICIP 2018

International audience; Human action recognition is a challenging task due to the complexity of human movements and to the variety among the same actions performed by distinct subjects. Recent technologies provide the skeletal representation of human body extracted in real time from depth maps, which is a high dis-criminant information for efficient action recognition. In this context, we present a new framework for human action recognition from skeleton sequences. We propose extracting sets of spatial and temporal local features from subgroups of joints, which are aggregated by a robust method based on the VLAD algorithm and a pool of clusters. Several feature vectors are then combined by a metric learning method inspired by the LMNN algorithm with the objective to improve the classification accuracy using the nonparametric k-NN classifier. We evaluated our method on three public datasets, including the MSR-Action3D, the UTKinect-Action3D, and the Florence 3D Actions dataset. As a result, the proposed framework performance overcomes the methods in the state of the art on all the experiments.

Human pose estimation and action recognition are related tasks since both problems are strongly dependent on the human body representation and analysis. Nonetheless, most recent methods in the literature handle the two problems separately. In this work, we propose a multi-task framework for jointly estimating 2D or 3D human poses from monocular color images and classifying human actions from video sequences. We show that a single architecture can be used to solve both problems in an efficient way and still achieves state-of-the-art or comparable results at each task while running at more than 100 frames per second. The proposed method benefits from high parameters sharing between the two tasks by unifying still images and video clips processing in a single pipeline, allowing the model to be trained with data from different categories simultaneously and in a seamlessly way. Additionally, we provide important insights for end-to-end training the proposed multi-task model by decoupling key prediction parts, which consistently leads to better accuracy on both tasks. The reported results on four datasets (MPII, Human3.6M, Penn Action and NTU RGB+D) demonstrate the effectiveness of our method on the targeted tasks. Our source code and trained weights are publicly available at https://github.com/dluvizon/deephar., Comment: Accepted to TPAMI. arXiv admin note: text overlap with arXiv:1802.09232

HighlightsÂ? New features for image search and object categorization. Â? Linearization of Kernels on Bags. Â? Derivative of second order statistics. Â? Linear/dot product metric. In this paper, we introduce a novel image signature effective in both image retrieval and image classification. Our approach is based on the aggregation of tensor products of discriminant local features, named VLATs (vector of locally aggregated tensors). We also introduce techniques for the packing and the fast comparison of VLATs. We present connections between VLAT and methods like kernel on bags and Fisher vectors. Finally, we show the ability of our method to be effective for two different retrieval problems, thanks to experiments carried out on similarity search and classification datasets.

International audience; Wove paper, made on a papermaking screen or mold having a surface of smooth tightly woven wires, was the predominant paper type used for printing in the twentieth century. To aid in the study and classification of fine art prints on wove paper, the present work compares the results of five different image processing approaches for characterizing surface texture. Using a collection of popular wove papers, a reference dataset of raking light close-up images was assembled. Five research teams undertook their own processing strategies to detect affinities among the paper samples. Their success in identifying similarity groupings are reported.

International audience; In this paper, we investigate the retrieval of paper watermark by visual similarity. We propose to perform the visual similarity by encoding small regions of the watermark using a non-negative dictionary optimized on a large collection of watermarks. The local codes are then aggregated into a single vector representing the whole watermark. Experiments are carried out on a test of tracings (manual binarization of watermarks).

International audience; In this paper, we propose an aggregation scheme of local descriptors that preserves local spatial information. Our method is based on the binary product of similarities of nearby matching pairs of descriptors. The similarities are linearized using a tensor framework. We show our approach can be used with any local descriptors, handcrafted like SIFT, or learned like the outputs of convolutional layers in deep neural networks. We perform experiments on the Holidays dataset that show the soundness of the approach.

There is something beguiling, something seductive, about these words of passion written by thirteenth-century Mechthild of Magdeburg (c. 1260-1294). These are not necessarily, or not only, words of love, but a testament to a series of mystical experiences in which the speaker is captivated, swept off her feet. Mechthild is not simply attracted to her lover; she is drawn to him in such an all-encompassing, insatiable way that her thoughts and actions are entirely consumed with the object of her passion, leaving her memories and attachments to everything else to fall away. This, she tells us in a later stanza, is incredibly painful: “I am wounded to death by your fiery love … and my senses ask me again and again/what is wrong with me” (1998: 94). Her body is powerless to stop her urges, and her lover, for his part, seems to relish in the chase. Stanzas alternate between her voice and his; he switches between affirming their attachment to one another, pledging love and protection, and withdrawing from her for her own good. His actions are strategic, designed to prolong-and indeed to grow-her obsession to the fullest. He holds the power; she almost seems powerless. This is no ordinary lover; he is a seducer.

International audience; In this article, we present a framework for distributed content based image retrieval with online learning based on ant-like mobile agents. Mobile agents crawl the network to find images matching a given example query. The images retrieved are shown to the user who labels them, following the classical relevant feedback scheme. The labels are used both to improve the similarity measure used for the retrieval and to learn paths leading to sites containing relevant images. The relevant paths are learned in an ethologically inspired way. We made experiments on the trecvid 2005 keyframe dataset showing that learning both the similarity function and the localization of the relevant images leads to a significant improvement. We also present an extension with the reuse of learned paths for later sessions leading to a further improvement.

International audience; This work takes place in the framework of GEODIFF project (funded by CNRS) and deals with the general issue of the social behavior modeling of pest insects with a particular focus on Bark Beetles. Bark Beetles are responsible for pine trees devastation in North America since 2005. In order to stem the problem and to apply an adapted strategy, one should be able to predict the evolution of the population of Bark Beetles. More precisely, a model taking into account a given population of insects (a colony) interacting with its environment, the forest ecosystem, would be very helpful. In a previous work, we aimed to model diffusive phenomenons across the environment using a simple reactive Multi-agent System. Bark beetle use pheromones as a support for recruitment of other bark beetles in the neighborhood in order to achieve a mass attack over a tree. They are first attracted by the ethanol or other phytopheromones emitted by a sick, stressed or dead tree and reinforce the presence of other individuals amongst the targeted tree. Both ethanol and semiochemicals are transported through the forest thanks to the wind, thermic effects and this advection phenomenon is modulated by the topology of the environment, tree and other obstacles distribution. In other words, the environment is involved in the process of a bark beetle attack. The first modeling we used to tackle our objective was not spatially explicit as long as free space propagation only was taken into account (isotropic phenomenon) with no constraint imposed by the environment such as wind. This article is intended to take into account such physical phenomenons and push the modeling one step further by providing predictions driven by measures provided by a Geographical Information System.

Local descriptors are the ground layer of recognition feature based systems for still images and video. We propose a new framework for the design of local descriptors and their evaluation. This framework is based on the descriptors decomposition in three levels: primitive extraction, primitive coding and code aggregation. With this framework, we are able to explain most of the popular descriptors in the literature such as HOG, HOF or SURF. This framework provides an efficient and rigorous approach for the evaluation of local descriptors, and allows us to uncover the best parameters for each descriptor family. Moreover, we are able to extend usual descriptors by changing the code aggregation or adding new primitive coding method. The experiments are carried out on images (VOC 2007) and videos datasets (KTH, Hollywood2, UCF11 and UCF101), and achieve equal or better performances than the literature. Author-HighlightsWe propose a new framework to design local visual descriptors.Descriptors are decomposed in primitive extraction, coding and aggregation.Our framework can explain the most popular descriptors such as HOG, HOF, SURF.The framework allows us a rigorous exploration of the possible combinations of primitives, coding and aggregation.New descriptors are easily obtained by changing one of the steps of our framework.

Ecotourism has become the focal point of a wide range of private and public sector development initiatives in the countries of the Western Indian Ocean. The 2008 “Regional Strategic Action Plan for Coastal Ecotourism Development in the South Western Indian Ocean” aimed at both a strategic assessment of current achievements and the development of recommendations for policy action. Drawing on the research data initially gathered for this report, the article demonstrates the importance of fostering criteria for ensuring medium and long-term economic viability in the assessment of sustainable ecotourism development performance, both at micro- and macro-economic levels. The arguments are presented in the form of eight lessons learnt, and the article approaches the question of economic viability at the different levels of enterprises, projects and products, and in terms of public policy and planning.

International audience; This paper investigates the classification of photographic paper tex-tures using visual descriptors. Such classification is called fine grain due to the very low inter-class variability. We propose a novel image representation for photographic paper texture categorization, relying on the incorporation of a powerful local descriptor into an efficient higher-order model deviation where texture is represented by com-puting statistics on the occurrences of specific local visual patterns. We perform an evaluation on two different challenging datasets of photographic paper textures and show such advanced methods in-deed outperforms existing descriptors.

International audience; Visual learning with weak supervision is a promising re-search area, since it offers the possibility to build large image datasets at reasonable cost. In this paper, we address the prob-lem of weakly supervised object detection, where the goal is to predict the label of the image using object position as latent variable. We propose a new method that builds upon the La-tent Structural SVM (LSSVM) formalism. Specifically, we introduce an original coarse-to-fine approach that limits the evolution of the latent parameter subspace. This incremental strategy drives the learning towards better solutions, provid-ing a model with increased predictive accuracy. In addition, this leads to a significant speed up during learning and infer-ence compared to standard sliding window methods. Experi-ments carried out on Mammal dataset validate the good per-formances and fast training of the method compared to state-of-the-art works.

International audience; In this paper, we propose a new method for taking into ac-count the spatial information in image categorization. More specifically, we remove the loss of spatial information in Bag of Words related methods by computing the image signature over specific regions selected by object detectors. We propose to select the detectors using Multiple Kernel Learning tech-niques. We carry out experiments on the well known VOC 2007 dataset, and show our semantic pooling obtains promis-ing results.

In this paper, we tackle the storage and computational cost of linear projections used in dimensionality reduction for near duplicate image retrieval. We propose a new method based on metric learning with a lower training cost than existing methods. Moreover, by adding a sparsity constraint, we obtain a projection matrix with a low storage and projection cost. We carry out experiments on a well known near duplicate image dataset and show our algorithm behaves correctly. Retrieval performances are shown to be promising when compared to the memory footprint and the projection cost of the obtained sparse matrix.

International audience; Several descriptors have been proposed in the past for 3D shape analysis, yet none of them achieves best perfor- mance on all shape classes. In this paper we propose a novel method for 3D shape analysis using the covariance matrices of the descriptors rather than the descriptors them- selves. Covariance matrices enable efficient fusion of dif- ferent types of features and modalities. They capture, using the same representation, not only the geometric and the spa- tial properties of a shape region but also the correlation of these properties within the region. Covariance matrices, however, lie on the manifold of Symmetric Positive Defi- nite (SPD) tensors, a special type of Riemannian manifolds, which makes comparison and clustering of such matrices challenging. In this paper we study covariance matrices in their native space and make use of geodesic distances on the manifold as a dissimilarity measure. We demonstrate the performance of this metric on 3D face matching and recognition tasks. We then generalize the Bag of Features paradigm, originally designed in Euclidean spaces, to the Riemannian manifold of SPD matrices. We propose a new clustering procedure that takes into account the geometry of the Riemannian manifold. We evaluate the performance of the proposed Bag of Covariance Matrices framework on 3D shape matching and retrieval applications and demonstrate its superiority compared to descriptor-based techniques.

International audience; This paper investigates the use of recent visual features based on second-order statistics, as well as new pro- cessing techniques to improve the quality of features. More specifically, we present and evaluate Fisher Vectors (FV), Vec- tors of Locally Aggregated Descriptors (VLAD), and Vectors of Locally Aggregated Tensors (VLAT). These techniques are combined with several normalization techniques, such as power law normalization and orthogonalisation/whitening of descriptor spaces. Results on the UC Merced land use dataset shows the relevance of these new methods for land-use classification, as well as a significant improvement over Bag-of-Words.

International audience; In this paper we propose to tackle human actions indexing by introducing a new local motion de-scriptor based on a model of the optical flow. We pro-pose to apply a coding step to vector field before the modeling. We use two modeling, a spatial model and a temporal model. The spatial model is computed by pro-jection of optical flow onto bivariate orthogonal poly-nomials. Then, the time evolution of spatial coefficients is modeled with a one dimension polynomial basis. To perform the action classification, we extend recent still image signatures using local descriptors to our proposal and combine them with linear SVM classifiers. The ex-periments are carried out on the well known UCF11 dataset and on the more challenging Hollywood2 ac-tion classification dataset and show promising results.

International audience; In web-scale image retrieval, the most effective strategy is to ag-gregate local descriptors into a high dimensionality signature and then reduce it to a small dimensionality. Thanks to this strategy, web-scale image databases can be represented with small index and explored using fast visual similarities. However, the computation of this index has a very high complexity, because of the high di-mensionality of signature projectors. In this work, we propose a new efficient method to greatly reduce the signature dimensionality with low computational and storage costs. Our method is based on the linear projection of the signature onto a small subspace using a sparse projection matrix. We report several experimental results on two standard datasets (Inria Holidays and Oxford) and with 100k image distractors. We show that our method reduces both the projec-tors storage cost and the computational cost of projection step while incurring a very slight loss in mAP (mean Average Precision) per-formance of these computed signatures.

IEEE International Conference on Image Processing 2013; International audience; In this paper, we present an efficient 3D object retrieval method invariant to scale, orientation and pose. Our approach is based on the dense extraction of discriminative local descriptors extracted from 2D views. We aggregate the descriptors into a single vector signature using tensor products. The similarity between 3D models can then be efficiently computed with a simple dot product. Experiments on the SHREC12 commonly-used benchmark demonstrate that our approach obtains superior performance in searching for generic shapes.

International audience; The main issues of web scale image retrieval are to achieve a good accuracy while retaining low computational time and memory footprint. In this paper, we propose a compact image signature by aggregating tensors of visual descriptors. Efficient aggregation is achieved by preprocessing the descriptors. Compactness is achieved by projection and quantization of the signatures. We compare our method to other efficient signatures on a 1 million images dataset, and show the soundness of the approach.

International audience; This paper addresses the challenging problem of scene classification in street-view georeferenced images of urban environments. More precisely, the goal of this task is semantic image classification, consisting in predicting in a given image, the presence or absence of a pre-defined class (e.g. shops, vegetation, etc.). The approach is based on the BOSSA representation, which enriches the Bag of Words (BoW) model, in conjunction with the Spatial Pyramid Matching scheme and kernel-based machine learning techniques. The proposed method handles problems that arise in large scale urban environments due to acquisition conditions (static and dynamic objects/pedestrians) combined with the continuous acquisition of data along the vehicle's direction, the varying light conditions and strong occlusions (due to the presence of trees, traffic signs, cars, etc.) giving rise to high intra-class variability. Experiments were conducted on a large dataset of high resolution images collected from two main avenues from the 12th district in Paris and the approach shows promising results.

International audience; Within the Content Based Image Retrieval (CBIR) frame- work, one of the main challenges is to tackle the scalability issues. We propose a new compact signature for similarity search. We use an original method to perform a high com- pression of signatures while retraining their effectiveness. We propose an embedding method that maps large signatures into a low-dimensional hilbert space. We evaluated the method on Holidays database and compared the results with methods of state-of-the-art.

International audience; In this article, we introduce an application of swarm intelligence to distributed visual information retrieval distributed over networks. Based on the relevance feedback scheme, we use ant-like agents to crawl the network and to retrieve relevant images. Agents movements are influenced by markers stored on the hosts. These markers are reinforced to match the distribution of relevant images over the network. We tackle the use of the information gathered during previous search sessions. In order to match the different categories available on the network, we use several markers. Sessions searching for the same category will thus use the same makers. The system involves three learning problems: the selection of relevant markers regarding the searched category, the reinforcement of these markers and the learning of the relevance function. All of these problems are based on the relevance feedback loop. We test our system on a custom network hosting images taken from the well known TrecVid dataset. Our system shows a high improvement over classical content based image retrieval systems which do not use previous sessions information.

International audience; Within the Content Based Image Retrieval (CBIR) framework, three main points can be highlighted: visual descriptors extraction, image signatures and their associated similarity measures, and machine learning based relevance functions. While the first and the last points have vastly improved in recent years, this paper addresses the second point. We propose a novel approach to compute vector representations extending state of the art methods in the field. Furthermore, our method can be viewed as a linearization of efficient well known kernel methods. The evaluation shows that our representation significantly improve state of the art results on the difficult VOC2007 database by a fair margin.

The efficacy of acupuncture in smoking cessation, and its effect on the urge to smoke are unclear. We evaluated the effect of a standardized protocol of transcutaneous electric acupoint stimulations (TEAS) on alleviating the urge to smoke. Ninety-eight smokers were recruited in two double-blind studies. Participants abstained from smoking for 26 h, and were randomized to receive TEAS alternating between 2 and 100 Hz at four acupoints (LI4 and PC8, PC6 and TE5) at four different intensities (10, 5, Intermittent 5 or 0 mA). The urge to smoke was assessed by the Questionnaire of Smoking Urges (QSU-Brief). In Experiment 1, the 10 mA group (n= 20) was compared with the 5 mA group (n= 20); the increase in smoking urges did not differ significantly. Considering the possibility that 5 mA may be an active intervention, in Experiment 2, a true placebo (0 mA), and a proxy of placebo [Intermittent 5 mA (i5 mA)] were compared with 10 mA TEAS. In this experiment, 10 mA (n= 20) TEAS showed a tendency to alleviate smoking urges compared with 0 mA (n= 16), and i5 mA (n= 19) TEAS. Only when the data of smokers with Fagerstöm Test for Nicotine Dependence score ≥5 were analyzed that the difference between the 10 mA group and the control group (0 and i5 mA) became significant. Based on these preliminary findings, we conclude that TEAS applied on the skin may antagonize the increase in urge to smoke in abstinent-dependent smokers. It seems warranted to assess the efficacy of TEAS in smoking cessation clinical trials involving a larger population of dependent smokers.

International audience; Recently, increasing interest has been brought to improve image categorization performances by combining multiple descriptors. However, very few approaches have been proposed for combining features based on complementary aspects, and evaluating the performances in realistic databases. In this paper, we tackle the problem of combining different feature types (edge and color), and evaluate the performance gain in the very challenging VOC 2009 benchmark. Our contribution is three-fold. First, we propose new local color descriptors, unifying edge and color feature extraction into the "Bag Of Word" model. Second, we improve the Spatial Pyramid Matching (SPM) scheme for better incorporating spatial information into the similarity measurement. Last but not least, we propose a new combination strategy based on ℓ1 Multiple Kernel Learning (MKL) that simultaneously learns individual kernel parameters and the kernel combination. Experiments prove the relevance of the proposed approach, which outperforms baseline combination methods while being computationally effective.

International audience; The count intercept is a robust method for the numerical analysis of fabrics Launeau and Robin (1996). It counts the number of intersections between a set of parallel scan lines and a mineral phase, which must be identified on a digital image. However, the method is only sensitive to boundaries and therefore supposes the user has some knowledge about their significance. The aim of this paper is to show that a proper grey level detection of boundaries along scan lines is sufficient to calculate the two-dimensional anisotropy of grain or crystal distributions without any particular image processing. Populations of grains and crystals usually display elliptical anisotropies in rocks. When confirmed by the intercept analysis, a combination of a minimum of 3 mean length intercept roses, taken on 3 more or less perpendicular sections, allows the calculation of 3-dimensional ellipsoids and the determination of their standard deviation with direction and intensity in 3 dimensions as well. The feasibility of this quick method is attested by numerous examples on theoretical objects deformed by active and passive deformation, on BSE images of synthetic magma flow, on drawing or direct analysis of thin section pictures of sandstones and on digital images of granites directly taken and measured in the field.

Background and Purpose— Stroke is a leading cause of death and disability worldwide. Despite improvements in acute stroke treatment, many patients only make a partial or poor recovery. Therefore, there is a need for treatments that would further improve outcome. Danqi Piantang Jiaonang (DJ; NeuroAid), a traditional Chinese medicine widely used in China to improve recovery after stroke, has been compared with another traditional Chinese medicine in 2 unpublished randomized clinical trials. The results of these studies were pooled and reanalyzed to assess efficacy and safety. Methods— Six hundred five subjects were randomized in 2 randomized double-blinded, controlled trials to receive either DJ or Buchang Naoxintong Jiaonang. Subjects were treated for 1 month. Inclusion criteria were: (1) patients with recent (from 10 days to 6 months) ischemic stroke; (2) patients satisfying Western diagnostic standards for stroke and traditional Chinese medicine standards for diagnosis of apoplexy; and (3) Diagnostic Therapeutic Effects of Apoplexy score ≥10. Results— The functional outcome, measured by the Comprehensive Function Score component of the Diagnostic Therapeutic Effects of Apoplexy scale, showed a statistically significant superiority of DJ over the control treatment group (relative risk, 2.4; 95% CI, 1.28 to 4.51; P =0.007). Tolerance was excellent in both groups. Conclusions— The pooled analysis of 2 unpublished trials of DJ, a traditional Chinese medicine currently approved in China to improve neurological recovery after stroke, shows good tolerability and superiority of DJ over another traditional Chinese medicine also approved for stroke. A large double-blind randomized clinical trial is required to further assess the safety and efficacy of DJ.

International audience; In this paper, we present a long term learning system for con- tent based image retrieval over a network. Relevant feedback is used among different sessions to learn both the similarity function and the best routing for the searched category. Our system is based on mobile agents crawling the network in search of relevant images. An ant-behavior algorithm is used to learn the category dependent routing. With experiments on trecvid'05 key-frame dataset, we show that the smart associ- ation of category dependent routing and active learning leads to an improvement of the quality of the retrieval.

International audience; Image retrieval over a network is the focus of this paper, as being a major challenge of content based image retrieval. We present a system that gathers feedbacks given by the users in order to learn the location of the searched images. As a result, the active learning of a content based relevance function is enhanced by the selection of hosts containing relevant examples. We achieve a long term merging of feedbacks over many sessions, without any knowledge about the category being searched at any session. Our system is based on mobile agents crawling the network in an ant-like behavior. Markers are used to learn a routing of the agent leading them to the relevant images. The benefits of the ant-like agents system is a natural parallelization of the processing as well as a distributed approach to the routing learning. We made experiments on the trecvidpsila05 key-frame dataset showing that the location of the categories were efficiently learned. Furthermore, the long-term learning of categories improves the interaction by reducing the number of labels needed during the interaction to obtain satisfying results.

Drawing on ethnographic data collected in a coastal village in southwestern Madagascar, the work explores the concept of memory through a study of contemporary hospitality practices aimed towards western tourists and other outsiders. The key observation is that memory is shaped by the interactions between two forms of colonial legacy. One is defined by the socially inscribed memory of past relations with powerful outsiders, mediated through myth and oral history. The other is defined by socially incorporated practices formed in the pre-colonial and colonial realms of the 19th and 20th centuries (and hence constitute a memory of such pasts in the present), and that today shift in the context of contact with western tourists and other foreigners. The main argument is that socially inscribed memory provides a highly flexible operational metaphor to mediate and also reproduce the socially incorporated practice-memory of hospitality, especially when historical symbolic functions - for example domesticating royals, colonial administrators, settlers, merchants or clerics - are shifted into contemporary contact zones with western tourists.

A volatility tandem differential mobility analyzer (VTDMA) was developed to allow fast field measurement of the volatile fraction of atmospheric aerosol particles in the particle size range 20–500 nm. In this VTDMA the volatile compounds are evaporated by heating the aerosol to a temperature between 25°C and 300°C. The heating unit is equipped with six symmetric columns kept at different temperatures that allow the heating temperature to be rapidly changed so that a higher temporal resolution can be achieved compared to a regular VTDMA. This work first focuses on the design and calibration of the heating units for the conditioning of a selected aerosol sample while minimizing sample losses due to thermophoresis and diffusion. The design was based on the modeling of the profiles of temperature and velocity and the behavior of a monodisperse aerosol in the heating units, using Computational Fluid Dynamics (CFD) Flow Modeling Software. This allowed for initial estimations of the heater dimensions and also ca...