Charco, J. L., Sappa, A.D., Vintimilla, B.X., Velesaca, H.O. (2021). Camera pose estimation in multi-view environments:from virtual scenarios to the real world. In Image and Vision Computing Journal. (Article number 104182), Vol. 110.
Abstract: This paper presents a domain adaptation strategy to efficiently train network architectures for estimating the relative camera pose in multi-view scenarios. The network architectures are fed by a pair of simultaneously acquired
images, hence in order to improve the accuracy of the solutions, and due to the lack of large datasets with pairs of
overlapped images, a domain adaptation strategy is proposed. The domain adaptation strategy consists on transferring the knowledge learned from synthetic images to real-world scenarios. For this, the networks are firstly
trained using pairs of synthetic images, which are captured at the same time by a pair of cameras in a virtual environment; and then, the learned weights of the networks are transferred to the real-world case, where the networks are retrained with a few real images. Different virtual 3D scenarios are generated to evaluate the
relationship between the accuracy on the result and the similarity between virtual and real scenarios—similarity
on both geometry of the objects contained in the scene as well as relative pose between camera and objects in the
scene. Experimental results and comparisons are provided showing that the accuracy of all the evaluated networks for estimating the camera pose improves when the proposed domain adaptation strategy is used,
highlighting the importance on the similarity between virtual-real scenarios.
|
|
Rosero Vasquez Shendry. (2020). Facial recognition: traditional methods vs. methods based on deep learning. Advances in Intelligent Systems and Computing – Information Technology and Systems Proceedings of ICITS 2020.615–625.
|
|
Patricia L. Suarez. (2020). Procesamiento y representación de imágenes multiespectrales usando técnicas de aprendizaje profundo (Ph.D. Angel Sappa, Director & Ph.D. Boris Vintimilla, Codirector.). Ph.D. thesis. In Ediciones FIEC-ESPOL..
|
|
Morocho-Cayamcela, M. E. (2020). Increasing the Segmentation Accuracy of Aerial Images with Dilated Spatial Pyramid Pooling. Electronic Letters on Computer Vision and Image Analysis (ELCVIA), Vol. 19(Issue 2), pp. 17–21.
|
Morocho-Cayamcela, M. E. & W. L. (2020). Lateral confinement of high-impedance surface-waves through reinforcement learning. Electronics Letters, Vol. 56(23, 12 November 2020), pp. 1262–1264.
Abstract: The authors present a model-free policy-based reinforcement learning
model that introduces perturbations on the pattern of a metasurface.
The objective is to learn a policy that changes the size of the
patches, and therefore the impedance in the sides of an artificially structured
material. The proposed iterative model assigns the highest reward
when the patch sizes allow the transmission along a constrained path
and penalties when the patch sizes make the surface wave radiate to
the sides of the metamaterial. After convergence, the proposed
model learns an optimal patch pattern that achieves lateral confinement
along the metasurface. Simulation results show that the proposed
learned-pattern can effectively guide the electromagnetic wave
through a metasurface, maintaining its instantaneous eigenstate when
the homogeneity is perturbed. Moreover, the pattern learned to
prevent reflections by changing the patch sizes adiabatically. The
reflection coefficient S1, 2 shows that most of the power gets transferred
from the source to the destination with the proposed design.
|
|
Patricia L. Suárez, A. D. S. and B. X. V. (2021). Deep learning-based vegetation index estimation. In Generative Adversarial Networks for Image-to-Image Translation Book. (Vol. Chapter 9, pp. 205–232).
|
Ángel Morera, Á. S., A. Belén Moreno, Angel D. Sappa, & José F. Vélez. (2020). SSD vs. YOLO for Detection of Outdoor Urban Advertising Panels under Multiple Variabilities. In Sensors, Vol. 2020-August(16), pp. 1–23.
Abstract: This work compares Single Shot MultiBox Detector (SSD) and You Only Look Once (YOLO)
deep neural networks for the outdoor advertisement panel detection problem by handling multiple
and combined variabilities in the scenes. Publicity panel detection in images o ers important
advantages both in the real world as well as in the virtual one. For example, applications like Google
Street View can be used for Internet publicity and when detecting these ads panels in images, it could
be possible to replace the publicity appearing inside the panels by another from a funding company.
In our experiments, both SSD and YOLO detectors have produced acceptable results under variable
sizes of panels, illumination conditions, viewing perspectives, partial occlusion of panels, complex
background and multiple panels in scenes. Due to the diculty of finding annotated images for the
considered problem, we created our own dataset for conducting the experiments. The major strength
of the SSD model was the almost elimination of False Positive (FP) cases, situation that is preferable
when the publicity contained inside the panel is analyzed after detecting them. On the other side,
YOLO produced better panel localization results detecting a higher number of True Positive (TP)
panels with a higher accuracy. Finally, a comparison of the two analyzed object detection models
with di erent types of semantic segmentation networks and using the same evaluation metrics is
also included.
|
Cristhian A. Aguilera, C. A., Cristóbal A. Navarro, & Angel D. Sappa. (2020). Fast CNN Stereo Depth Estimation through Embedded GPU Devices. Sensors 2020, Vol. 2020-June(11), pp. 1–13.
Abstract: Current CNN-based stereo depth estimation models can barely run under real-time
constraints on embedded graphic processing unit (GPU) devices. Moreover, state-of-the-art
evaluations usually do not consider model optimization techniques, being that it is unknown what is
the current potential on embedded GPU devices. In this work, we evaluate two state-of-the-art models
on three different embedded GPU devices, with and without optimization methods, presenting
performance results that illustrate the actual capabilities of embedded GPU devices for stereo depth
estimation. More importantly, based on our evaluation, we propose the use of a U-Net like architecture
for postprocessing the cost-volume, instead of a typical sequence of 3D convolutions, drastically
augmenting the runtime speed of current models. In our experiments, we achieve real-time inference
speed, in the range of 5–32 ms, for 1216 368 input stereo images on the Jetson TX2, Jetson Xavier,
and Jetson Nano embedded devices.
|
Nayeth I. Solorzano Alcivar, R. L., Stalyn Gonzabay Yagual, & Boris X. Vintimilla. (2020). Statistical Representations of a Dashboard to Monitor Educational Videogames in Natural Language. In ETLTC – ACM Chapter: International Conference on Educational Technology, Language and Technical Communication; Fukushima, Japan, 27-31 Enero 2020 (Vol. 77).
Abstract: This paper explains how Natural Language (NL) processing by computers, through smart
programs as a way of Machine Learning (ML), can represent large sets of quantitative data as written
statements. The study recognized the need to improve the implemented web platform using a
dashboard in which we collected a set of extensive data to measure assessment factors of using
children´s educational games. In this case, applying NL is a strategy to give assessments, build, and
display more precise written statements to enhance the understanding of children´s gaming behavior.
We propose the development of a new tool to assess the use of written explanations rather than a
statistical representation of feedback information for the comprehension of parents and teachers with
a lack of primary level knowledge in statistics. Applying fuzzy logic theory, we present verbatim
explanations of children´s behavior playing educational videogames as NL interpretation instead of
statistical representations. An educational series of digital game applications for mobile devices,
identified as MIDI (Spanish acronym of “Interactive Didactic Multimedia for Children”) linked to a
dashboard in the cloud, is evaluated using the dashboard metrics. MIDI games tested in local primary
schools helps to evaluate the results of using the proposed tool. The guiding results allow analyzing
the degrees of playability and usability factors obtained from the data produced when children play a
MIDI game. The results obtained are presented in a comprehensive guiding evaluation report
applying NL for parents and teachers. These guiding evaluations are useful to enhance children's
learning understanding related to the school curricula applied to ludic digital games.
|
Marjorie Chalen, & Boris X. Vintimilla. (2019). Towards Action Prediction Applying Deep Learning. Latin American Conference on Computational Intelligence (LA-CCI); Guayaquil, Ecuador; 11-15 Noviembre 2019, , pp. 1–3.
Abstract: Considering the incremental development future action prediction by video analysis task of computer vision where it is done based upon incomplete action executions. Deep learning is playing an important role in this task framework. Thus, this paper describes recently techniques and pertinent datasets utilized in human action prediction task.
|
