<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Computer Vision |</title><link>https://www.ugr.es/~cvrlab/tags/computer-vision/</link><atom:link href="https://www.ugr.es/~cvrlab/tags/computer-vision/index.xml" rel="self" type="application/rss+xml"/><description>Computer Vision</description><generator>HugoBlox Kit (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Thu, 19 Jan 2023 00:00:00 +0000</lastBuildDate><image><url>https://www.ugr.es/~cvrlab/media/icon_hu_138e5d72c01fd19a.png</url><title>Computer Vision</title><link>https://www.ugr.es/~cvrlab/tags/computer-vision/</link></image><item><title>BRAINAV - BRAIn-inspired visual processing for real-time energy-efficient autonomous NAVigation</title><link>https://www.ugr.es/~cvrlab/projects/brainav/</link><pubDate>Thu, 19 Jan 2023 00:00:00 +0000</pubDate><guid>https://www.ugr.es/~cvrlab/projects/brainav/</guid><description>&lt;ul&gt;
&lt;li&gt;This work was supported by the National Project PID2022-141466OB-I00 funded by MICIU/AEI/10.13039/501100011033 and by ERDF/EU.
&lt;figure&gt;&lt;img src="https://www.ugr.es/~cvrlab/projects/brainav/aei_logo.png"&gt;
&lt;/figure&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="summary"&gt;Summary&lt;/h2&gt;
&lt;p&gt;BRAINAV aims at developing bio-inspired pipelines for real-time energy-efficient visual navigation that operates in the real-world in optimized platforms embedded onboard mobile robots. First, we need the development of a new paradigm for computer vision that allows us to effectively work with asynchronous neuromorphic sensors that imitate visual retinas and the development of high-performance architectures for reaching real-time performance with them. Then, we also require efficient ML networks for reaching low-energy inference and eventually, to propose new bio-inspired components to solve mapping, localization, and path planning for exploration. We are committed to prove the potential of our energy-efficient vision pipelines and therefore, we propose the use case of autonomous navigation using mobile robots.&lt;/p&gt;
&lt;p&gt;Video processing is crucial in perception due for environment understanding. However, computer vision is a very demanding task and in robotics it requires to close perception-action loops in real-time. Moreover, distributed local mobile nodes need low-energy operation since they run on batteries. Power consumption vs. performance is a challenging trade-off, especially in distributed environments, given the number of sensors, the heterogeneity of the system, their autonomy or their need of high performance. Our hypothesis is that biologically-inspired strategies will help us achieve energy-efficient solutions that reach good levels of accuracy while operating in real-time in real scenarios.&lt;/p&gt;
&lt;p&gt;The final demonstrators consist of UAV platforms equipped with event and conventional cameras, inertial sensors, telemetry and GNSS, with one configuration additionally including LiDAR. All perception, planning and control components run onboard using an NVIDIA Jetson Orin. Real outdoor experiments validated autonomous flight, trajectory tracking at speeds up to 8 m/s, detection of other UAVs, autonomous obstacle perception and avoidance, and simple LiDAR-based mapping. BRAINAV contributes directly to the thematic priority &amp;ldquo;Digital World, Industry, Space and Defence&amp;rdquo; through advances in artificial intelligence, autonomous robotics, neuromorphic vision and embedded processing.&lt;/p&gt;
&lt;h2 id="objectives"&gt;Objectives&lt;/h2&gt;
&lt;p&gt;The main general objective of BRAINAV is the development of a bio-inspired pipeline for real-time
energy-efficient visual navigation that operates in the real-world in optimized platforms embedded in
mobile robots. First, we need the development of a new paradigm for computer vision that allows us
to effectively work with these asynchronous neuromorphic sensors and the development of high-
performance architectures for reaching real-time performance. Then, we also require efficient ML
networks for reaching low-energy inference and eventually, to propose new bio-inspired components
to solve mapping, localization, and path planning for exploration. Finally, we also propose to prove
the potential of our energy-efficient vision pipelines for autonomous navigation using mobile robots.&lt;/p&gt;
&lt;p&gt;Our specific objectives (SO) are detailed next:&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO1. To develop of event-driven perception for navigation tasks.&lt;/em&gt;
As mentioned before in the state of the art, some works have addressed the task previously with event-
based sensors [BAR14a, BAR21, VID18]. However, their results are still very poor, reaching only low
performance. In our previous work, we have thoroughly studied alternative knowledge-based methods
for 3D perception, mainly image motion and 3D motion estimation. In BRAINAV, we
propose to study new methods based on Machine Learning solutions [DEN23, ISE23] to achieve
better accuracy performance. Also, most of the previous methods build images out of accumulation of
events for prefixed time intervals and apply classic methods to these artificial images. This severely
harms the performance and does not take advantage of the main feature of neuromorphic sensors, their
high-temporal resolution or low latency. We propose the development and evaluation of event-driven
methods for motion and 3D geometry estimation [BAR21]. Lastly, the accuracy of our solution will be
evaluated using novel benchmarks and datasets that integrate both events and images using the motion
capture system available in our lab. Moreover, we are committed to the Open Science principles and
will release the datasets for the community. This is crucial for the adv ancement of neuromorphic
engineering, due the current lack of datasets and their relevance for Machine Learning techniques.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO2. To optimize for real-time energy-efficient processing engines.&lt;/em&gt;
In an effort to reduce the computational complexity of the prop osed solutions, we will explore
different mechanisms for low-energy computation. First, we propose the use of optimization
techniques on ANNs such as: model quantization [LIA21] shortening bit widths for inference to reduce
computational complexity; pruning [VAD22], that removes connections with smaller values in
networks; and distillation [DEN23], that allows transferring knowledge from teacher complex models
to student simpler ones. Second, we are also interested in exploring new techniques based on Spiking
Neural Networks. These are biologically-plausible neural architectures do take into account energy
consumption [DAV21] doing learning with sparse data. We also propose the exploration of brain-like
structures to build new navigation methods that imitate their functionalities.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO3. To implement real-time processing on embedded platforms with neuromorphic sensors.&lt;/em&gt;
Dealing with a real-world application imposes restrictions such as real-time performance for closing
our perception-action loops for autonomous navigation. We will consider different hardware
accelerators such as GPUs or FPGAs, using the latest advancements for the development of fine-
tuning pipelines (e.g. DeepStream) and for optimized hardware datapaths (e.g. Vitis AI). The selected
hardware accelerator will impose different non-functional requirements on the implementations
developed for SO1 and functionally optimized for SO2, for performance, latency and power
consumption. This objective also pursues the effective integration of the asynchrono us sensors into
von Neumann architectures which is by itself, an additional challenge.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO4. To demonstrate embedded autonomous navigation onboard mobile robots&lt;/em&gt;
Finally, this objective aims at the validation of the proposed models and architectures considered in
the SO1-SO3, with different demonstrators for autonomous navigation tasks in real-world
environments. This application allows us to demonstrate the maximum potential of our optimized bio-
inspired pipelines and downstream processing, while ensuring the overall optimization in the
exploration task relying mainly on learning models. Our goal in SO4 is to develop a first semi-
autonomous platform to perform first localization and mapping. This prototype will have a minimum
intervention of a human pilot, and will be used to validate the mentioned tasks in unknown scenarios.
The second prototype will show the final autonomous explorer that will be enabled with additional
downstream applications such as the detection of other mobile robots in the scenario and possibly
tracking them, or finding specific targets.&lt;/p&gt;
&lt;h2 id="final-autonomous-navigation-demonstrator"&gt;Final autonomous navigation demonstrator&lt;/h2&gt;
&lt;p&gt;The final BRAINAV demonstrator integrates the main scientific and technological results of the project into a real autonomous UAV. The platform combines:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Neuromorphic event cameras&lt;/li&gt;
&lt;li&gt;Conventional cameras&lt;/li&gt;
&lt;li&gt;Inertial sensing&lt;/li&gt;
&lt;li&gt;GNSS and telemetry&lt;/li&gt;
&lt;li&gt;LiDAR in one of the configurations&lt;/li&gt;
&lt;li&gt;NVIDIA Jetson Orin onboard processing&lt;/li&gt;
&lt;li&gt;ROS-based perception, planning and control&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;All computation is performed onboard the UAV. The platform was validated through real outdoor flights at the Cubillas test area.&lt;/p&gt;
&lt;figure&gt;&lt;img src="https://www.ugr.es/~cvrlab/projects/brainav/brainav_flight_test.jpg"
alt="UAV platform and outdoor autonomous flight experiments at the Cubillas test area."&gt;&lt;figcaption&gt;
&lt;p&gt;UAV platform and outdoor autonomous flight experiments at the Cubillas test area.&lt;/p&gt;
&lt;/figcaption&gt;
&lt;/figure&gt;
&lt;h2 id="background-references"&gt;Background references&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;[BAR14a] F. Barranco, C. Fermüller, Y. Aloimonos. &amp;ldquo;Contour motion estimation for asynchronous event-driven cameras.&amp;rdquo; Proceedings of the IEEE 102, no. 10 (2014): 1537-1556.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[BAR21] F. Barranco, C. Fermüller, Y. Aloimonos, E. Ros. &amp;ldquo;Joint direct estimation of 3D geometry and 3D motion using spatio temporal gradients.&amp;rdquo; Pattern Recognition, 113, 2021.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[DAV21] S. Davidson, S. B. Furber. &amp;ldquo;Comparison of artificial and spiking neural networks on digital hardware.&amp;rdquo; Frontiers in Neuroscience 15 (2021): 651141.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[DEN23] D. Deniz, J. Isern, J. Solanti, P. Jääskeläinen, P. Hnětynka, L. Bulej, E. Ros, F. Barranco. &amp;ldquo;Efficient reconfigurable CPS for monitoring the elderly at home via Deep Learning.&amp;rdquo; Journal of Signal Processing Systems, (2023).&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[ISE23] J. Isern, G. Jimenez-Perera, L. Medina-Valdés, P. Chaves, D.Pampliega F. Ramos, F. Barranco. &amp;ldquo;A Cyber-Physical System for integrated remote control and protection of smart grid critical infrastructures.&amp;rdquo;, Journal of Signal Processing Systems, 2023.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[LIA21] T. Liang, J. Glossner, L. Wang, S. Shi, X. Zhang. &amp;ldquo;Pruning and quantization for deep neural network acceleration: A survey.&amp;rdquo; Neurocomputing 461 (2021): 370 -403.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[VAD22] S. Vadera, S. Ameen. &amp;ldquo;Methods for pruning deep neural networks.&amp;rdquo; IEEE Access 10 (2022): 63280-63300.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[VID18] A. Vidal, H. Rebecq, T. Horstschaefer, D. Scaramuzza. &amp;ldquo;Ultimate SLAM? Combining events, images, and IMU for robust visual SLAM.&amp;rdquo; IEEE Robotics and Automation 3, no. 2 (2018).&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="lists-of-contributions-publications-datasets-and-code-repositories"&gt;Lists of contributions: publications, datasets, and code repositories&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;[DEN24] Deniz, D., Ros, E., Ortigosa, E. M., &amp;amp; Barranco, F. (2024). &lt;strong&gt;Optimized edge-cloud system for activity monitoring using knowledge distillation.&lt;/strong&gt; Electronics, 13(23), 4786.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[NOV24] Novo, A., Lobon, F., Garcia de Marina, H., Romero, S., &amp;amp; Barranco, F. (2024). &lt;strong&gt;Neuromorphic perception and navigation for mobile robots: a review.&lt;/strong&gt; ACM Computing Surveys, 56(10), 1-37.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[DEN25] Deniz, D., Isern, J., Solanti, J., Jääskeläinen, P., Hnětynka, P., Bulej, L., Ros, E., &amp;amp; Barranco, F. (2025). &lt;strong&gt;Efficient reconfigurable system for home monitoring of the elderly via action recognition.&lt;/strong&gt; Engineering Applications of Artificial Intelligence, 158, 111383.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[JIM25] Jimenez-Perera, G., Valencia-Vidal, B., Luque, N. R., Ros, E., &amp;amp; Barranco, F. (2025). &lt;strong&gt;Informed Federated Learning to Train a Robotic Arm Inverse Dynamic Model.&lt;/strong&gt; IEEE Robotics and Automation Letters, 10(10), 11022-11029.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[VAZ26] Vázquez, V., Valenzuela, E., Shepstone, R., Megías, C., Miccichè, G., Ros, E., &amp;amp; Barranco, F. (2026). &lt;strong&gt;Remote handling operation for IFMIF-DONES supported by time-sensitive networking.&lt;/strong&gt; Nuclear Fusion, 66, 046029.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[NOV25a] Novo, A., Fermüller, C., Rodríguez-Álvarez, M., Romero, S., &amp;amp; Barranco, F. (2026). &lt;strong&gt;On ego-motion estimation from event-inertial-aided normal flow fields.&lt;/strong&gt; Computer Vision and Image Understanding. Under review.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[NOV25b] Novo, A., Fermüller, C., Romero, S., Rodríguez-Álvarez, M., &amp;amp; Barranco, F. (2026). &lt;strong&gt;Event-based vision for online inference of fine-grained manipulation actions.&lt;/strong&gt; Engineering Applications of Artificial Intelligence. Under review (second round).&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="other-scientific-contributions"&gt;Other scientific contributions&lt;/h3&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;[SAL25-CONG] Salinas, I., Romero, S. F., Molero, J. A., Gutiérrez de León, B., Banqueri, J., &amp;amp; Barranco, F. (2025). &lt;strong&gt;Clasificación de UAS basado en Inteligencia Artificial para dar soporte a sistemas de defensa antiaérea.&lt;/strong&gt; XII Congreso Nacional de I+D en Defensa y Seguridad, Zamora, Spain, pp. 105-106. ISBN 978-84-1083-073-8. Premio Antonio Torres al mejor trabajo (ISDEFE).&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[HER25] Heredia, A. J., Pérez, F. V., Novo, A., Romero, S. F., García Imbernón, G., Ortega, J. F., García, V., Albaladejo, D., Méndez, M., &amp;amp; Barranco, F. (2025). &lt;strong&gt;Automatización inteligente del sostenimiento en buques mediante sensórica avanzada e IA.&lt;/strong&gt; XII Congreso Nacional de I+D en Defensa y Seguridad, Zamora, Spain, pp. 167-168. ISBN 978-84-1083-073-8.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;[CUA23] Cuadrado, J., Barranco, F., &amp;amp; Masquelier, T. (2023). &lt;strong&gt;Optical flow estimation from event- and frame-based camera fusion.&lt;/strong&gt; Biocomp Colloque 2023, Toulouse, France.&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="datasets-and-open-source-resources"&gt;Datasets and open-source resources&lt;/h3&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;IndoorActionDataset&lt;/strong&gt; — Dataset associated with [DEN24].
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;CIAV Dataset&lt;/strong&gt; — Dataset associated with [DEN25].
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Efficient action recognition models&lt;/strong&gt; — Code and neural-network weights associated with [DEN25].
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Baxter robot dataset for learning dynamic models&lt;/strong&gt; — Dataset associated with [JIM25], released under CC BY 4.0.
&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;E-MAD — Event-based Manipulation Action Dataset&lt;/strong&gt; — Dataset associated with [NOV25b]. The dataset is currently under embargo during the publication process and will be released under CC BY 4.0 after publication.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Event-inertial ego-motion code&lt;/strong&gt; — Code associated with [NOV25a]. The repository will be made public after publication of the corresponding article.&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="authors"&gt;Authors&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Francisco Barranco&lt;/strong&gt; - Principal Investigator, University of Granada, CVRLab / CITIC-UGR&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="acknowledgments"&gt;Acknowledgments&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;This work was supported by the National Project PID2022-141466OB-I00 funded by MICIU/AEI/10.13039/501100011033 and by ERDF/EU.
&lt;figure&gt;&lt;img src="https://www.ugr.es/~cvrlab/projects/brainav/aei_logo.png"&gt;
&lt;/figure&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="license"&gt;License&lt;/h2&gt;
&lt;p&gt;Software, datasets and other research outputs associated with BRAINAV are distributed under the licenses indicated in their corresponding repositories.&lt;/p&gt;
&lt;p&gt;Unless otherwise stated, the contents of this website are provided for scientific dissemination purposes. Please refer to each publication, dataset or software repository for its specific copyright and licensing conditions.&lt;/p&gt;</description></item><item><title>BIO-PERCEPTION - Next generation of smart vision systems for real-time processing with bio-inspired sensors</title><link>https://www.ugr.es/~cvrlab/projects/bioperception/</link><pubDate>Mon, 20 Dec 2021 00:00:00 +0000</pubDate><guid>https://www.ugr.es/~cvrlab/projects/bioperception/</guid><description>&lt;ul&gt;
&lt;li&gt;This work was supported by the AEI through the grant PID2019-109434RA-I00 / AEI / 10.13039/501100011033.
&lt;figure&gt;&lt;img src="https://www.ugr.es/~cvrlab/projects/bioperception/aei_logo.jpeg"&gt;
&lt;/figure&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="summary"&gt;Summary&lt;/h2&gt;
&lt;p&gt;The 3D perception of our environment such as the estimation of image motion, 3D motion, or the 3D structure of unknown scenes, are tasks that humans efficiently do. In fact, we do it in a precise manner enabling us to navigate around our environment and interact with complex dynamic scenarios in a natural way. Despite of the advances in robotics, computational capacity, and computer vision, the current state of the art cannot provide efficient solutions that are even close in performance. It is consequence of the impossibility of closing perception-action cycles. Every action performed causes a change in the environment that the perception system needs to sense, to implement the next action accordingly. The key factor for the success of the current smart vision systems is not only the perception, or the execution of the correct action; it is the capacity of performing the whole loop in real-time to allow the continuous interaction. Conventional solutions propose the increase of the computational complexity, without taking into account factors such as the energy consumption, required resources, or processing time. The most advanced high-performance architectures have problems dealing with the amount of information generated by high-resolution cameras fast enough for enabling dynamic agents to interact in the real world (for example, state- of-the-art solutions take minutes for image motion estimation). Moreover, processing has to be local in order to respond to the real-time
demands.&lt;/p&gt;
&lt;p&gt;Our goal for this project is precisely to set up the basis for a new generation of smart autonomous agents that are able to carry out 3D perception in real-time, using biologically-inspired vision sensors. These sensors are based on the manner visual information is processed by the human eye. Rather than transmitting all the static information continuously, the retina selects the most relevant information, comprising it to be further processed in the visual cortex. In the very same way, these sensors independently processed all pixels and only trigger changes in the scene (&amp;rsquo;events&amp;rsquo;) in the case a substantial difference in the intensity luminance happens over time for a specific location (this happens only at object contours and textures). This allows for the reduction of the transmission of redundant information and hence, the data bandwidth. The other key advantage is that these sensors reach very low latencies, of a few microseconds.&lt;/p&gt;
&lt;p&gt;The current paradigm for computation with this kind of sensors is still very basic. Usually, asynchronous events are accumulated during short prefixed time intervals to create synthetic images and then, classic methods are applied to them. Unfortunately, this causes the increase of latencies and it generates redundant information that has to be processed in the same manner systems for conventional sensors do. We propose in this project a novel paradigm for event-driven computation: 1) event-wise processing, for every event that comes in, the model is updated and the event information is integrated for a specific task, taking full advantage of the high-temporal
resolution and 2) only the relevant information is transmitted.&lt;/p&gt;
&lt;p&gt;Finally, regarding the resources we will use hardware accelerators for massively parallel processing such as GPU or FPGAs that process huge amounts of data at once, guaranteeing bounded latencies and low energy consumption. This will also enable the integration of our system via embedded platforms.&lt;/p&gt;
&lt;h2 id="objectives"&gt;Objectives&lt;/h2&gt;
&lt;p&gt;BIO-PERCEPTION is focused on the development of a new paradigm for vision systems that integrates neuromorphic event-based sensors. Additionally, we propose the validation of this new model by building smart vision systems for autonomous navigation, proposing drones for that. Our specific objectives (SO) are detailed next:&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO1. Development of a new model for event-based 3D perception.&lt;/em&gt;
As mentioned before in the state of the art, some works have addressed the task previously with event-based sensors [BEN12, BEN14, BAR14]. However, their results are still very poor, reaching only low performance. Moreover, most of these methods build images out of accumulation of events for prefixed time intervals and apply classic methods to these artificial images. This severely harms the performance and does not take advantage of the main feature of these sensors, their high-temporal resolution or low latency.
We propose the development and evaluation of a fully event-driven model, although it will also include the possibility of integrating conventional synchronous images to improve accuracy. The models for motion object contours will be applied here [FER01, BAR14], to estimate the image motion. Using the image motion, 3D pose and velocity can also be subsequently estimated [FER01, BAR16] and, independently moving objects (IMOs) detected in the scene. Lastly, the accuracy of our solution will be evaluated using novel benchmarks and datasets that integrate both events and images.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO2. Development of a new model for event-based object segmentation.&lt;/em&gt;
Although the 3D perception module provides sparse results, object segmentation is an image-based task, which implies the need for a centralized model of the scene. In SO2, our first goal is to come up with a linking mechanism to group events and be able to distinguish different depth layers (separated by contours). Our group has already developed some models such as in [BAR15b] that could not reach real-time performance or the required accuracy. Given the popularity of Machine Learning approaches and the expertise of our Department in efficient machine learning applications, we will consider the use of Deep Learning, to develop architectures based on CNNs (Convolutional Neural Networks). These networks have proven their success in this task for images [SHE15]. After separating events from contours and events generated due to the textured regions of the objects in the scene, we will group them together in sets according to the object these events belong to. We also propose the implementation of an efficient mechanism to do that. As in SO1, we will produce new benchmarks and datasets to assess the accuracy of our models, and compare conventional and event-based approaches.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO3. Real-time processing architectures with event-based neuromorphic sensors.&lt;/em&gt;
In the motivation we already stated that autonomy cannot be reached until perception-action loops are closed in real-time. This means that the visual perception processing should be fast enough to allow agents to interact with the dynamic real-world. Moreover, in order to embed the visual processing in our autonomous agents, we should consider platforms to integrate our processing architectures on. In order to do that, we propose the use of hardware accelerators. Our group has long expertise on optimized-CPU implementations, and GPUs, or FPGAs [BAR14b, NAV17]. Also, we will consider integrated solutions with the existing UAV platforms and their embedded processors. Models and implementations designed according to SO1 and SO2 will be adapted to the non-functional requirements related to performance, latency, or memory imposed by the selected architecture.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;SO4. Demonstrator: Autonomous drone platform.&lt;/em&gt;
Finally, we also consider that the validation of the proposed models has to be shown in the framework of autonomous navigation tasks. This application allows us to demonstrate the maximum potential of the neuromorphic sensors, specifically regarding the high-temporal resolution and low latency, for example with high-speed maneuvers. Our goal in SO4 is to develop a first semi-autonomous platform that shows navigation with a minimum intervention of a human pilot for the first case, and that validates tasks such as selection of targets, segmentation, or detection of independently moving objects while navigating. The second prototype will show the whole system working to do tasks completely autonomously, without human intervention, while navigating avoiding obstacles.&lt;/p&gt;
&lt;h2 id="bibliography"&gt;Bibliography&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;[BAR14a] F. Barranco, C. Fermüller, Y. Aloimonos. &amp;ldquo;Contour motion estimation for asynchronous event-driven cameras.&amp;rdquo; Proceedings of the IEEE 102, no. 10 (2014): 1537-1556.&lt;/li&gt;
&lt;li&gt;[BAR14b] F. Barranco, J. Diaz, B. Pino, E. Ros. &amp;ldquo;Real-time visual saliency architecture for FPGA with top-down attention modulation.&amp;rdquo; IEEET on Ind. Informatics 10, 3, 726-1735, 2014.&lt;/li&gt;
&lt;li&gt;[BAR15] F. Barranco, C. Fermuller, Y. Aloimonos. &amp;ldquo;Bio-inspired motion estimation with event-driven sensors.&amp;rdquo; In International Work-Conference on Artificial Neural Networks, pp. 309-321, 2015.&lt;/li&gt;
&lt;li&gt;[BAR16] F. Barranco, C. Fermuller, Y. Aloimonos, T. Delbruck. &amp;ldquo;A dataset for visual navigation with neuromorphic methods.&amp;rdquo; Frontiers in neuroscience 10 (2016): 49.&lt;/li&gt;
&lt;li&gt;[BEN12] R. Benosman, S. Ieng, C. Clercq, C. Bartolozzi, M. Srinivasan. &amp;ldquo;Asynchronous frameless event-based optical flow.&amp;rdquo; Neural Networks 27 (2012): 32-37.&lt;/li&gt;
&lt;li&gt;[BEN14] R. Benosman, C. Clercq, X. Lagorce, S. Ieng, C. Bartolozzi. &amp;ldquo;Event-based visual flow.&amp;rdquo; IEEE transactions on neural networks and learning systems 25, no. 2 (2014): 407-417.&lt;/li&gt;
&lt;li&gt;[FER01] C. Fermüller, D. Shulman, Y. Aloimonos. &amp;ldquo;The statistics of optical flow.&amp;rdquo; Computer Vision and Image Understanding 82, no. 1 (2001): 1-32.&lt;/li&gt;
&lt;li&gt;[NAV17] F. Naveros, J. Garrido, R. Carrillo, E. Ros, N. Luque. &amp;ldquo;Event-and time-driven techniques using parallel CPU-GPU co-processing for spiking neural networks.&amp;rdquo; Front. in Neuroinf. 11 (2017): 7.&lt;/li&gt;
&lt;li&gt;[SHE15] W. Shen, X. Wang, Y. Wang, X. Bai, Z. Zhang. &amp;ldquo;Deepcontour: A deep convolutional feature learned by positive-sharing loss for contour detection.&amp;rdquo; CVPR, pp. 3982-3991. 2015.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="lists-of-contributions-publications-datasets-and-code-repositories"&gt;Lists of contributions: publications, datasets, and code repositories&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;Isern, J., F. Barranco, D. Deniz, J. Lesonen, J. Hannuksela, R. R. Carrillo. &lt;strong&gt;Reconfigurable cyber-physical system for critical infrastructure protection in smart cities via smart video-surveillance.&lt;/strong&gt; Pattern Recognition Letters 140 (2020): 303-309.
&lt;/li&gt;
&lt;li&gt;Déniz, D., G. Jimenez-Perera, R. Nolasco, J. Corral, F. Barranco. &lt;strong&gt;Deep multimodal habit tracking system: A user-adaptive approach for low-power embedded systems.&lt;/strong&gt; Journal of Signal Processing Systems (2023),
&lt;/li&gt;
&lt;li&gt;Isern, J., G. Jimenez-Perera, L. Medina-Valdés, P. Chaves, D. Pampliega, F. Ramos, F. Barranco. &lt;strong&gt;A Cyber-Physical System for integrated remote control and protection of smart grid critical infrastructures.&lt;/strong&gt; Journal of Signal Processing Systems (2023),
&lt;/li&gt;
&lt;li&gt;Deniz, D., J. Isern, J. Solanti, P. Jääskeläinen, P. Hnětynka, L. Bulej, E. Ros, F. Barranco. &lt;strong&gt;Efficient reconfigurable CPS for monitoring the elderly at home via Deep Learning&lt;/strong&gt;, submitted to Engineering Applications of Artificial Intelligence, 2022.
&lt;/li&gt;
&lt;li&gt;Cuadrado, J., U. Rancon, B.R. Cottereau, F. Barranco, T. Masquelier, &lt;strong&gt;Optical flow estimation from event-based cameras and spiking neural networks&lt;/strong&gt;, Frontiers in Neuroscience, 17, 1160034, (2023)
&lt;/li&gt;
&lt;li&gt;Deniz, D., C. Fermuller, E. Ros, M.Rodriguez-Alvarez, F. Barranco. &lt;strong&gt;Event-based vision for early prediction of manipulation actions&lt;/strong&gt;, submitted since January 2023 to IEEE T. on Pattern analysis and machine intelligence.
&lt;/li&gt;
&lt;li&gt;Deniz, D., E. Ros, E. M. Ortigosa, F. Barranco. &lt;strong&gt;Optimized edge-cloud system for activity monitoring using knowledge distillation&lt;/strong&gt;, submitted June 2023 to IEEE T. on Mobile Computing.&lt;/li&gt;
&lt;li&gt;Novo, A., F. Lobon, H. G. de Marina, S. Romero, F. Barranco. &lt;strong&gt;Brain-Inspired Autonomous Robotic Navigation: A Survey&lt;/strong&gt;, submitted in March 2023 to ACM Computing Surveys.&lt;/li&gt;
&lt;li&gt;Barranco, F., M. Rodríguez-Álvarez, S. Romero, E. Martínez-Ortigosa. &lt;strong&gt;Learning object contours using bio-inspired spatio-temporal features&lt;/strong&gt;, In preparation, 2023.&lt;/li&gt;
&lt;li&gt;Lobón, F., A. Novo, S. Romero, F. Barranco. &lt;strong&gt;Segmentation of Independently-Moving Objects using neuromorphic sensors&lt;/strong&gt;, In preparation, 2023.
&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="authors"&gt;Authors&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Francisco Barranco&lt;/strong&gt; - &lt;em&gt;University of Granada&lt;/em&gt;
Please report problems, bugs, or suggestions to fbarranco_at_ugr_dot_es (Replace &lt;em&gt;at&lt;/em&gt; by @ and &lt;em&gt;dot&lt;/em&gt; by .).&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="acknowledgments"&gt;Acknowledgments&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;This work was supported by the AEI through the grant PID2019-109434RA-I00 / AEI / 10.13039/501100011033.&lt;/li&gt;
&lt;/ul&gt;
&lt;figure&gt;&lt;img src="https://www.ugr.es/~cvrlab/projects/bioperception/aei_logo.jpeg"&gt;
&lt;/figure&gt;
&lt;h2 id="license"&gt;License&lt;/h2&gt;
&lt;p&gt;Copyright (C) 2018 Francisco Barranco, 01/09/2018, University of Granada.&lt;/p&gt;
&lt;p&gt;This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License.&lt;/p&gt;
&lt;p&gt;This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.&lt;/p&gt;
&lt;p&gt;You should have received a copy of the GNU General Public License along with this program. If not, see
.&lt;/p&gt;</description></item></channel></rss>