Browsing by Author "Neves, João"
Now showing 1 - 6 of 6
Results Per Page
Sort Options
- Automatic annotation of cellular dataPublication . Neves, João; Proença, Hugo Pedro Martins CarriçoLife scientists often need to count cells in microscopy images, which is very tedious and a time consuming task. Henceforth, automatic approaches can be a solution to this problem. Several works have been devised for this issue, but the majority of these approaches degrade their performance in case of cell overlapping. In this dissertation we propose a method to determine the position of macrophages and parasites in uorescence images of Leishmania-infected macrophages. The proposed strategy is mainly based on blob detection, clustering and separation using concave regions of the cells' contour. By carrying out a comparison with other approaches that also addressed this type of images, we concluded that the proposed methodology achieves better performance in the automatic annotation of Leishmania infections.
- Deep-PRWIS: Periocular Recognition Without the Iris and Sclera Using Deep Learning FrameworksPublication . Proença, H.; Neves, JoãoThis work is based on a disruptive hypothesisfor periocular biometrics: in visible-light data, the recognitionperformance is optimized when the components inside the ocularglobe (the iris and the sclera) are simply discarded, and therecogniser’s response is exclusively based in information fromthe surroundings of the eye. As major novelty, we describe aprocessing chain based on convolution neural networks (CNNs)that defines the regions-of-interest in the input data that should beprivileged in an implicit way, i.e., without masking out any areasin the learning/test samples. By using an ocular segmentationalgorithm exclusively in the learning data, we separate the ocularfrom the periocular parts. Then, we produce a large set of”multi-class” artificial samples, by interchanging the periocularand ocular parts from different subjects. These samples areused for data augmentation purposes and feed the learningphase of the CNN, always considering as label the ID of theperiocular part. This way, for every periocular region, the CNNreceives multiple samples of different ocular classes, forcing itto conclude that such regions should not be considered in itsresponse. During the test phase, samples are provided withoutany segmentation mask and the networknaturallydisregardsthe ocular components, which contributes for improvements inperformance. Our experiments were carried out in full versionsof two widely known data sets (UBIRIS.v2 and FRGC) and showthat the proposed method consistently advances the state-of-the-art performance in theclosed-worldsetting, reducing the EERsin about 82% (UBIRIS.v2) and 85% (FRGC) and improving theRank-1 over 41% (UBIRIS.v2) and 12% (FRGC).
- Fusing Vantage Point Trees and Linear Discriminants for Fast Feature ClassificationPublication . Proença, H.; Neves, JoãoThis paper describes a classification strategy that can be regarded as amore general form of nearest-neighbor classification. It fuses the concepts ofnearestneighbor,linear discriminantandVantage-Pointtrees, yielding an efficient indexingdata structure and classification algorithm. In the learning phase, we define a set ofdisjoint subspaces of reduced complexity that can be separated by linear discrimi-nants, ending up with an ensemble of simple (weak) classifiers that work locally. Inclassification, the closest centroids to the query determine the set of classifiers con-sidered, which responses are weighted. The algorithm was experimentally validatedin datasets widely used in the field, attaining error rates that are favorably compara-ble to the state-of-the-art classification techniques. Lastly, the proposed solution hasa set of interesting properties for a broad range of applications: 1) it is determinis-tic; 2) it classifies in time approximately logarithmic with respect to the size of thelearning set, being far more efficient than nearest neighbor classification in terms ofcomputational cost; and 3) it keeps the generalization ability of simple models.
- IRINA: Iris Recognition (even) in Inacurately Segmented DataPublication . Proença, H.; Neves, JoãoThe effectiveness of current iris recognition systems de-pends on the accurate segmentation and parameterisationof the iris boundaries, as failures at this point misalignthe coefficients of the biometric signatures. This paper de-scribesIRINA, an algorithm forIrisRecognition that is ro-bust againstINAccurately segmented samples, which makesit a good candidate to work in poor-quality data. The pro-cess is based in the concept of ”corresponding” patch be-tween pairs of images, that is used to estimate the posteriorprobabilities that patches regard the same biological region,even in case of segmentation errors and non-linear texturedeformations. Such information enables to infer a free-formdeformation field (2D registration vectors) between images,whose first and second-order statistics provide effective bio-metric discriminating power. Extensive experiments werecarried out in four datasets (CASIA-IrisV3-Lamp, CASIA-IrisV4-Lamp, CASIA-IrisV4-Thousand and WVU) and showthat IRINA not only achieves state-of-the-art performancein good quality data, but also handles effectively severe seg-mentation errors and large differences in pupillary dilation/ constriction.
- A Leopard Cannot Change Its Spots: Improving Face Recognition Using 3D-based CaricaturesPublication . Neves, João; Proença, H.Caricatures refer to a representation of aperson in which the distinctive features are deliberatelyexaggerated, with several studies showing that humansperform better at recognizing people from caricaturesthan using original images. Inspired by this observa-tion, this paper introduces the first fully automatedcaricature-based face recognition approach capable ofworking with data acquired in the wild. Our approachleverages the 3D face structure from a single 2D imageand compares it to a reference model for obtaininga compact representation of face features deviations.This descriptor is subsequently deformed using a ’mea-sure locally, weight globally’ strategy to resemble thecaricature drawing process. The deformed deviationsare incorporated in the 3D model using the Laplacianmesh deformation algorithm, and the 2D face cari-cature image is obtained by projecting the deformedmodel in the original camera-view. To demonstratethe advantages of caricature-based face recognition, wetrain the VGG-Face network from scratch using eitheroriginal face images (baseline) or caricatured images,and use these models for extracting face descriptorsfrom the LFW, IJB-A and MegaFace datasets. The ex-periments show an increase in the recognition accuracywhen using caricatures rather than original images.Moreover, our approach achieves competitive resultswith state-of-the-art face recognition methods, evenwithout explicitly tuning the network for any of theevaluation sets.
- Mechanical Properties of Fused Filament Fabrication MaterialsPublication . Santos, João; Silva, Jorge; Gamboa, Pedro; Neves, JoãoThe last decades have witnessed a significant development and increased usage of additive manufacturing (AM) technologies. Among these technologies, the fused filament fabrication (FFF) is being characterized by a massive worldwide spreading, from specialized companies with high-quality equipment to the hobbyist with home-made 3D printers. Thus, and aiming to help the user achieving the best results, it is essential to continuously perform studies and analyses concerning the mechanical properties resulting from different feedstock materials and printing settings. This study covers the execution and the subsequent comparative analysis of the results of flexural and tensile tests on three different types of FFF materials: polylactic acid (PLA), acrylonitrile butadiene styrene (ABS), and CarbonFilTM. It also comprehends a parallel study of the infill setting, more specifically the testing of specimens with a 50% infill reduction. All tests were executed according to the corresponding American Society for Testing and Materials (ASTM) standards. This work aims to clarify the properties of three different materials and the effect of the internal material reduction on the mechanical properties of 3D printed objects.