159 research outputs found
Automatic estimation of the number of deformation modes in non-rigid SfM with missing data
No full textThis paper proposes a new algorithm to estimate automatically the number of deformation modes needed to describe a non-rigid object with the well-known low-rank shape model, focusing on the missing data case. The 3D shape is assumed to deform as a linear combination of K rigid shape bases according to time varying coefficients. One of the requirements of this formulation is that the number of bases must be known in advance. Most non-rigid structure from motion (NRSfM) approaches based on this model determine the value of K empirically. Our proposed approach is based on the analysis of the frequency spectra of the x and y coordinates corresponding to the individual image trajectories, which are seen as 1D signals. The frequency content of the 2D trajectories is encoded using the modulus of the Discrete Cosine Transform (DCT) of the signals. Our hypothesis is that the value of K that gives the best prediction of the missing data also provides the best 3D reconstruction. Our proposed approach does not assume any prior knowledge and is independent of the 3D reconstruction algorithm used. We validate our approach with experiments on synthetic and real sequences.Carme Julià, Marco Paladini, Ravi Garg, Domenec Puig, and Lourdes Agapit
An embodied-simplexity approach to design humanoid robots bioinspired by taekwondo athletes
No full textWe are investigating new approaches to design and develop embodied humanoid robots with predictive behaviour in the real world. Our approaches are strongly based on the symbiotic interaction between the concepts of the embodied intelligence and simplexity that enables us to reproduce the artificial body in symbiosis with its intelligence. The research is based on the study of martial arts athletes and their planned movements. In particular, the taekwondo martial art competition where anticipation and adaptive behaviours are key points is used as a vehicle to address conceptual and design issues. The expected outcome will be the development of a robot that will have the capability to anticipate an opponent's actions by coordinating eyes, head and legs, with a stable body constituted by an embodied platform bioinspired by the taekwondo athletes. © 2014 Springer-Verlag
Developing biorobotics for veterinary research into cat movements
No full textCollaboration between veterinarians and other professionals such as engineers and computer scientists will become important in biorobotics for both scientific achievements and the protection of animal welfare. Particularly, cats have not yet become a significant source of inspiration for new technologies in robotics. This article suggests a novel approach for the investigation of particular aspects of cat morphology, neurophysiology, and behavior aimed at bridging this gap by focusing on the versatile, powerful locomotion abilities of cats and implementing a robotic tool for the measurements of biological parameters of animals and building cat-inspired robotic prototypes. The presented framework suggests the basis for the development of novel hypotheses and models describing biomechanics, locomotion, balancing system, visual perception, as well as learning and adaption of cat motor skills and behavior. In subsequent work, the resulting models will be tested and evaluated in simulated and real experiments and validated with specific experimental data gathered from cats. This methodology has application in several areas including dynamic models and artificial vision systems. From an ethical point of view, this approach is in line with the 3R principles: the detailed and integrated systems will allow us to study a small number of cats (reduction) for the implementation of noninvasive tools such as electromyography and gaze analysis (refinement), which will make the construction of a substitute to experiments on living cats (replacement) easier. For instance, bioinspired prototypes could be used to test how specific visual and physical impairment in cats (up to partial or total blindness, loss of a leg, and so forth) change their walking and jumping abilities. This modus operandi may pave the way for a new generation of research in the veterinary field. Moreover, the measurement tools to be developed will constitute an achievement per se as for the first time visual, muscular, and gait analysis of cats will be integrated, and this will help to improve the rehabilitation procedures for cats and other nonhuman animals
VISTA: vision improvement via split and reconstruct deep neural network for fundus image quality assessment
No full textWidespread eye conditions such as cataracts, diabetic retinopathy, and glaucoma impact people worldwide. Ophthalmology uses fundus photography for diagnosing these retinal disorders, but fundus images are prone to image quality challenges. Accurate diagnosis hinges on high-quality fundus images. Therefore, there is a need for image quality assessment methods to evaluate fundus images before diagnosis. Consequently, this paper introduces a deep learning model tailored for fundus images that supports large images. Our division method centres on preserving the original image’s high-resolution features while maintaining low computing and high accuracy. The proposed approach encompasses two fundamental components: an autoencoder model for input image reconstruction and image classification to classify the image quality based on the latent features extracted by the autoencoder, all performed at the original image size, without alteration, before reassembly for decoding networks. Through post hoc interpretability methods, we verified that our model focuses on key elements of fundus image quality. Additionally, an intrinsic interpretability module has been designed into the network that allows decomposing class scores into underlying concepts quality such as brightness or presence of anatomical structures. Experimental results in our model with EyeQ, a fundus image dataset with three categories (Good, Usable, and Rejected) demonstrate that our approach produces competitive outcomes compared to other deep learning-based methods with an overall accuracy of 0.9066, a precision of 0.8843, a recall of 0.8905, and an impressive F1-score of 0.8868.Fil: Khalid, Saif. Universitat Rovira I Virgili; EspañaFil: Abdulwahab, Saddam. Universitat Rovira I Virgili; EspañaFil: Stanchi, Oscar Agustín. Universidad Nacional de La Plata. Facultad de Informática. Instituto de Investigación en Informática Lidi; Argentina. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - La Plata; ArgentinaFil: Quiroga, Facundo Manuel. Universidad Nacional de La Plata. Facultad de Informática. Instituto de Investigación en Informática Lidi; Argentina. Provincia de Buenos Aires. Gobernación. Comisión de Investigaciones Científicas; ArgentinaFil: Ronchetti, Franco. Provincia de Buenos Aires. Gobernación. Comisión de Investigaciones Científicas; Argentina. Universidad Nacional de La Plata. Facultad de Informática. Instituto de Investigación en Informática Lidi; ArgentinaFil: Puig, Domenec. Universitat Rovira I Virgili; EspañaFil: Rashwan, Hatem A.. Universitat Rovira I Virgili; Españ
A personal robotic flying machine with vertical takeoff controlled by the human body movements
No full textWe propose a cooperative research project aimed at designing and prototyping a new generation of personal flying robotic platform controlled by movements of the human body using a symbiotic human-robot-flight machine interaction. Motors with ducted fun propulsion and power supply, and a VSLAM system will be integrated in the final flight machine with short and vertical takeoff and landing capability and composite (or light alloy) airframe structure for low speed and low altitude flight. In the project, we will also develop a flight simulator to test the interaction between the flying machine and the human body movements. In this first step, for human safety, the flying machine will be controlled by an autopilot colligated in a closed-loop control with the simulator
A personal robotic flying machine with vertical takeoff controlled by the human body movements
No full textWe propose a cooperative research project aimed at designing and prototyping a new generation of personal flying robotic platform controlled by movements of the human body using a symbiotic human-robot-flight machine interaction. Motors with ducted fun propulsion and power supply, and a VSLAM system will be integrated in the final flight machine with short and vertical takeoff and landing capability and composite (or light alloy) airframe structure for low speed and low altitude flight. In the project, we will also develop a flight simulator to test the interaction between the flying machine and the human body movements. In this first step, for human safety, the flying machine will be controlled by an autopilot colligated in a closed-loop control with the simulator. © 2014 Springer-Verlag
Convexity shape constraints for retinal blood vessel segmentation and foveal avascular zone detection
No full textDiabetic retinopathy (DR) has become a major worldwide health problem due to the increase in blindness among diabetics at early ages. The detection of DR pathologies such as microaneurysms, hemorrhages and exudates through advanced computational techniques is of utmost importance in patient health care. New computer vision techniques are needed to improve upon traditional screening of color fundus images. The segmentation of the entire anatomical structure of the retina is a crucial phase in detecting these pathologies. This work proposes a novel framework for fast and fully automatic blood vessel segmentation and fovea detection. The preprocessing method involved both contrast limited adaptive histogram equalization and the brightness preserving dynamic fuzzy histogram equalization algorithms to enhance image contrast and eliminate noise artifacts. Afterwards, the color spaces and their intrinsic components were examined to identify the most suitable color model to reveal the foreground pixels against the entire background. Several samples were then collected and used by the renowned convexity shape prior segmentation algorithm. The proposed methodology achieved an average vasculature segmentation accuracy exceeding 96%, 95%, 98% and 94% for the DRIVE, STARE, HRF and Messidor publicly available datasets, respectively. An additional validation step reached an average accuracy of 94.30% using an in-house dataset provided by the Hospital Sant Joan of Reus (Spain). Moreover, an outstanding detection accuracy of over 98% was achieved for the foveal avascular zone. An extensive state-of-the-art comparison was also conducted. The proposed approach can thus be integrated into daily clinical practice to assist medical experts in the diagnosis of DR.Escorcia-Gutierrez, Jose-will be generated-orcid-0000-0003-0518-3187-600Torrents-Barrena, Jordina-will be generated-orcid-0000-0002-7380-6297-600Gamarra, Margarita-will be generated-orcid-0000-0003-1834-2984-600Romero-Aroca, Pedro-will be generated-orcid-0000-0002-7061-8987-600Valls, Aida-will be generated-orcid-0000-0003-3616-7809-600Puig, Domenec-will be generated-orcid-0000-0002-0562-4205-60
- …
