Pasar al contenido principal

SENSORES CMOS DE VISION, GESTION DE ENERGIA Y SEGUIMIENTO DE OBJETOS SOBRE GPUS EMPOTRADAS

RTI2018-097088-B-C32

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Convocatoria Proyectos I+D+i Retos de Investigación
Año convocatoria 2018
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDADE DE SANTIAGO DE COMPOSTELA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 9
Encontrada(s) 1 página(s)

Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Ultra-Low-Power Low-Input-Voltage Charge Pump for Micro-Energy Harvesting Applications

  • Pereira Rial, Óscar
  • Cabrini, Alessandro
  • Torelli, Guido
  • López Martínez, Paula
  • Carrillo, Juan M.
A subthreshold input voltage charge pump based on the well-known cross-coupled voltage doubler and using boosted gate voltages for the transfer switches is presented. A level shifter and some inverters, including a novel inverter architecture proposed in this work and referred to as negative low-state voltage inverter, are used to generate the clock signals for the switching transistors with the purpose of significantly improving their drive capability. A complete analysis of the proposed charge pump is provided to highlight the advantages of the implemented structure, revealing the power efficiency improvement when the input voltage is below the threshold voltage of the transistors. An extensive experimental characterization of silicon prototypes in 180 nm CMOS technology was carried out, showing that the proposed scheme is able to pump charge from an input voltage as low as 110 mV. The experimental peak efficiency remains above 70% for input voltages between 180 mV and 400 mV and input power levels from 45 nW to 25 μW, which are appropriate for different miniaturized transducers implementable on chip.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

0.6-V-VIN 7.0-nA-IQ 0.75-mA-IL CMOS Capacitor-Less LDO for Low-Voltage Micro-Energy-Harvested Supplies

  • Pereira Rial, Óscar
  • López Martínez, Paula
  • Carrillo, Juan M.
A capacitor-less (CL) low-dropout (LDO) regulator suitable to be incorporated in an on-chip system with low-voltage micro-energy-harvested supply, is proposed in this contribution. The differential input stage of the error amplifier includes bulk-driven MOS transistors, thus providing the LDO with an output voltage range that extends from the negative rail up to a level very close to the input voltage without the need of using a resistive feedback network. The circuit parameters relying on the feedback factor, , are maximized thanks to the use of a unitary value for this parameter. The CL-LDO has been designed and fabricated in standard 180-nm CMOS technology and optimized to operate with an input voltage equal to 0.6 V and a reference level of 0.5 V. The experimental characterization of the fabricated prototypes shows that, under these operating conditions, the LDO is able to deliver a load current above 0.75 mA with a total quiescent current of only 7.0 nA. Furthermore, the proposed voltage regulator is able to operate from input voltages as low as 0.4 V, delivering in this case a maximum load current of 30 μA.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Short-term anchor linking and long-term self-guided attention for video object detection

  • Cores Costa, Daniel
  • Brea Sánchez, Víctor Manuel
  • Mucientes Molina, Manuel
We present a new network architecture able to take advantage of spatio-temporal information available in videos to boost object detection precision. First, box features are associated and aggregated by linking proposals that come from the same anchor box in the nearby frames. Then, we design a new attention module that aggregates short-term enhanced box features to exploit long-term spatio-temporal information. This module takes advantage of geometrical features in the long-term for the first time in the video object detection domain. Finally, a spatio-temporal double head is fed with both spatial information from the reference frame and the aggregated information that takes into account the short- and long-term temporal context. We have tested our proposal in five video object detection datasets with very different characteristics, in order to prove its robustness in a wide number of scenarios. Non-parametric statistical tests show that our approach outperforms the state-of-the-art. Our code is available at https://github.com/daniel-cores/SLTnet
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

On-Chip Solar Energy Harvester and PMU With Cold Start-Up and Regulated Output Voltage for Biomedical Applications

  • Cabello Ferrer, Diego
  • Ferro Santiago, Esteban
  • Pereira Rial, Óscar
  • Martínez Vázquez, Beatriz
  • Brea Sánchez, Víctor Manuel
  • Carrillo, Juan M.
This paper presents experimental results from a system that comprises a fully autonomous energy harvester with a solar cell of 1 mm 2 as energy transducer and a Power Management Unit (PMU) on the same silicon substrate, and an output voltage regulator. Both chips are implemented in standard 0.18 μm CMOS technology with total layout areas of 1.575 mm 2 and 0.0126 mm 2 , respectively. The system also contains an off-the-shelf 3.2 mm × 2.5 mm × 0.9 mm supercapacitor working as an off-chip battery or energy reservoir between the PMU and the voltage regulator. Experimental results show that the fast energy recovery of the on-chip solar cell and PMU permits the system to replenish the supercapacitor with enough charge as to sustain Bluetooth Low Energy (BLE) communications even with input light powers of 510 nW. The whole system is able to self-start-up without external mechanisms at 340 nW. This work is the first step towards a self-supplied sensor node with processing and communication capabilities. The small form factor and ultra-low power consumption of the system components is in compliance with biomedical applications requirements
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

A full data augmentation pipeline for small object detection based on generative adversarial networks

  • Bosquet Mera, Brais
  • Cores Costa, Daniel
  • Brea Sánchez, Víctor Manuel
  • Mucientes Molina, Manuel
  • Bimbo, Alberto del
Object detection accuracy on small objects, i.e., objects under 32 32 pixels, lags behind that of large ones. To address this issue, innovative architectures have been designed and new datasets have been released. Still, the number of small objects in many datasets does not suffice for training. The advent of the generative adversarial networks (GANs) opens up a new data augmentation possibility for training architectures without the costly task of annotating huge datasets for small objects. In this paper, we propose a full pipeline for data augmentation for small object detection which combines a GAN-based object generator with techniques of object segmentation, image inpainting, and image blending to achieve high-quality synthetic data. The main component of our pipeline is DS-GAN, a novel GAN-based architecture that generates realistic small objects from larger ones. Experimental results show that our overall data augmentation method improves the performance of state-of-the-art models up to 11.9% AP on UAVDT and by 4.7% AP on iSAID, both for the small objects subset and for a scenario where the number of training instances is limited.
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Real-time siamese multiple object tracker with enhanced proposals

  • Vaquero Otal, Lorenzo
  • Brea Sánchez, Víctor Manuel
  • Mucientes Molina, Manuel
Maintaining the identity of multiple objects in real-time video is a challenging task, as it is not always feasible to run a detector on every frame. Thus, motion estimation systems are often employed, which either do not scale well with the number of targets or produce features with limited semantic information. To solve the aforementioned problems and allow the tracking of dozens of arbitrary objects in real-time, we propose SiamMOTION. SiamMOTION includes a novel proposal engine that produces quality features through an attention mechanism and a region-of-interest extractor fed by an inertia module and powered by a feature pyramid network. Finally, the extracted tensors enter a comparison head that efficiently matches pairs of exemplars and search areas, generating quality predictions via a pairwise depthwise region proposal network and a multi-object penalization module. SiamMOTION has been validated on five public benchmarks, achieving leading performance against current state-of-the-art trackers. Code available at: https://www.github.com/lorenzovaquero/SiamMOTION
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

STDnet-ST: Spatio-temporal ConvNet for small object detection

  • Bosquet Mera, Brais
  • Mucientes Molina, Manuel
  • Brea Sánchez, Víctor Manuel
Object detection through convolutional neural networks is reaching unprecedented levels of precision. However, a detailed analysis of the results shows that the accuracy in the detection of small objects is still far from being satisfactory. A recent trend that will likely improve the overall object detection success is to use the spatial information operating alongside temporal video information. This paper introduces STDnet-ST, an end-to-end spatio-temporal convolutional neural network for small object detection in video. We define small as those objects under px, where the features become less distinctive. STDnet-ST is an architecture that detects small objects over time and correlates pairs of the top-ranked regions with the highest likelihood of containing those small objects. This permits to link the small objects across the time as tubelets. Furthermore, we propose a procedure to dismiss unprofitable object links in order to provide high quality tubelets, increasing the accuracy. STDnet-ST is evaluated on the publicly accessible USC-GRAD-STDdb, UAVDT and VisDrone2019-VID video datasets, where it achieves state-of-the-art results for small objects
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

Real-time visual detection and tracking system for traffic monitoring

  • Fernández Sanjurjo, Mauro
  • Bosquet Mera, Brais
  • Mucientes Molina, Manuel
  • Brea Sánchez, Víctor Manuel
Computer vision systems for traffic monitoring represent an essential tool for a broad range of traffic surveillance applications. Two of the most noteworthy challenges for these systems are the real-time operation with hundreds of vehicles and the total occlusions which hinder the tracking of the vehicles. In this paper, we present a traffic monitoring approach that deals with these two challenges based on three modules: detection, tracking and data association. First, vehicles are identified through a deep learning based detector. Second, tracking is performed with a combination of a Discriminative Correlation Filter and a Kalman Filter. This permits to estimate the tracking error in order to make tracking more robust and reliable. Finally, the data association through the Hungarian algorithm combines the information of the previous steps. The contributions are: (i) a real-time traffic monitoring system robust to occlusions that can process more than four hundred vehicles simultaneously; and (ii) the application of the system to anomaly detection in traffic and roundabout input/output analysis. The system has been evaluated with more than two thousand vehicles in real-life videos
Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela
Publicación Artículo científico (JournalArticle)

STDnet: Exploiting high resolution feature maps for small object detection

  • Bosquet Mera, Brais
  • Mucientes Molina, Manuel
  • Brea Sánchez, Víctor Manuel
The accuracy of small object detection with convolutional neural networks (ConvNets) lags behind that of larger objects. This can be observed in popular contests like MS COCO. This is in part caused by the lack of specific architectures and datasets with a sufficiently large number of small objects. Our work aims at these two issues. First, this paper introduces STDnet, a convolutional neural network focused on the detection of small objects that we defined as those under 16 × 16 pixels. The high performance of STDnet is built on a novel early visual attention mechanism, called Region Context Network (RCN), to choose the most promising regions, while discarding the rest of the input image. Processing only specific areas allows STDnet to keep high resolution feature maps in deeper layers providing low memory overhead and higher frame rates. High resolution feature maps were proved to be key to increasing localization accuracy in such small objects. Second, we also present USC-GRAD-STDdb, a video dataset with more than 56,000 annotated small objects in challenging scenarios. Experimental results over USC-GRAD-STDdb show that STDnet improves the AP@.5 of the best state-of-the-art object detectors for small target detection from 50.8% to 57.4%. Performance has also been tested in MS COCO for objects under 16 × 16 pixels. In addition, a spatio-temporal baseline network, STDnet-bST, has been proposed to make use of the information of successive frames, increasing the AP@.5 of STDnet in 2.3%. Finally, optimizations have been carried out to be fit on embedded devices such as Jetson TX2