scholarly journals RT-Seg: A Real-Time Semantic Segmentation Network for Side-Scan Sonar Images

Sensors ◽  
2019 ◽  
Vol 19 (9) ◽  
pp. 1985
Author(s):  
Qi Wang ◽  
Meihan Wu ◽  
Fei Yu ◽  
Chen Feng ◽  
Kaige Li ◽  
...  
Keyword(s):  
High Resolution ◽  
Real Time ◽  
Autonomous Underwater Vehicle ◽  
Semantic Segmentation ◽  
Processing Unit ◽  
Real Time Processing ◽  
Side Scan Sonar ◽  
Network Training ◽  
Sonar Sensor ◽  
Sonar Images

Real-time processing of high-resolution sonar images is of great significance for the autonomy and intelligence of autonomous underwater vehicle (AUV) in complex marine environments. In this paper, we propose a real-time semantic segmentation network termed RT-Seg for Side-Scan Sonar (SSS) images. The proposed architecture is based on a novel encoder-decoder structure, in which the encoder blocks utilized Depth-Wise Separable Convolution and a 2-way branch for improving performance, and a corresponding decoder network is implemented to restore the details of the targets, followed by a pixel-wise classification layer. Moreover, we use patch-wise strategy for splitting the high-resolution image into local patches and applying them to network training. The well-trained model is used for testing high-resolution SSS images produced by sonar sensor in an onboard Graphic Processing Unit (GPU). The experimental results show that RT-Seg can greatly reduce the number of parameters and floating point operations compared to other networks. It runs at 25.67 frames per second on an NVIDIA Jetson AGX Xavier on 500*500 inputs with excellent segmentation result. Further insights on the speed and accuracy trade-off are discussed in this paper.

Real-time, High-resolution Depth Upsampling on Embedded Accelerators

2021 ◽  
Vol 20 (3) ◽  
pp. 1-22
Author(s):  
David Langerman ◽  
Alan George
Keyword(s):  
High Resolution ◽  
Low Power ◽  
Real Time ◽  
Mixed Reality ◽  
Graphics Processing Unit ◽  
Processing Unit ◽  
Reconfigurable Logic ◽  
Depth Sensors ◽  
Time Requirements ◽  
Graphics Processing

High-resolution, low-latency apps in computer vision are ubiquitous in today’s world of mixed-reality devices. These innovations provide a platform that can leverage the improving technology of depth sensors and embedded accelerators to enable higher-resolution, lower-latency processing for 3D scenes using depth-upsampling algorithms. This research demonstrates that filter-based upsampling algorithms are feasible for mixed-reality apps using low-power hardware accelerators. The authors parallelized and evaluated a depth-upsampling algorithm on two different devices: a reconfigurable-logic FPGA embedded within a low-power SoC; and a fixed-logic embedded graphics processing unit. We demonstrate that both accelerators can meet the real-time requirements of 11 ms latency for mixed-reality apps. 1

scholarly journals Effective Parallelization Method for Object Recognition in 2D Sonar Images Based on Task Partitioning

2019 ◽  
Vol 2019 ◽  
pp. 1-9 ◽  
Author(s):  
Ok-Kyoon Ha ◽  
Keonpyo Lee ◽  
Wan-Jin Kim ◽  
Kun Su Yoon
Keyword(s):  
Object Recognition ◽  
Parallel Processing ◽  
Real Time ◽  
Processing Method ◽  
Task Partitioning ◽  
Real Time Processing ◽  
Sequential Processing ◽  
Sonar Images ◽  
Graphical Processing ◽  
Imaging Sonar

Techniques for analyzing and avoiding hazardous objects and situations on the seabed are being developed to ensure the safety of ships and submersibles from various hazards. Improvements in accuracy and real-time response are critical for underwater object recognition, which rely on underwater sonar detection to remove noises and analyze the data. Therefore, parallel processing is being introduced for real-time processing of two-dimensional (2D) underwater sonar detector images for seabed monitoring. However, this requires optimized parallel processing between the modules for image processing and the data processing of a vast amount of data. This study proposes an effective parallel processing method, called Task Partitioning, based on central and graphical processing units for monitoring and identifying underwater objects in real time based on 2D-imaging sonar. The practicality of the proposed method is evaluated experimentally by comparing it to the sequential processing method. The experimental results show that the Task Partitioning method significantly improves the processing time for sonar images because it reduces the average execution time to 1% and 5% of the sequential processing method and general parallelization, respectively.

Side Scan Sonar Modeling for Maritime Robotic Operations

2012 ◽  
Vol 152-154 ◽  
pp. 1195-1201
Author(s):  
Kuan Meng Tan ◽  
Tien Fu Lu ◽  
Amir Anvar
Keyword(s):  
Autonomous Underwater Vehicle ◽  
Underwater Vehicle ◽  
Simulation Framework ◽  
Test Bed ◽  
Side Scan Sonar ◽  
Sensor Model ◽  
Sonar Sensor ◽  
Key Aspects ◽  
The University ◽  
Sensor Modeling

One of the key aspects in designing an Autonomous Underwater Vehicle (AUV) simulation framework is sensor modeling. This paper presents specifically the underwater sonar sensor modeling structure used in the proposed AUV simulation framework. This sensor model covers the mathematical aspects from the field of acoustics which mimics real world sensors. Simplified sonar signal models are widely used however rarely discussed in the literature. Based on this designed simulation framework, simple scenario using different sonar configuration is shown and discussed. This paper shows the formulation of a typical side-scan sonar with emphasis on the assumptions which leads to the simplification of the sonar model. The sonar sensor model is built based on a developed AUV test-bed which was done previously in the University of Adelaide.

scholarly journals Real-Time Moving Object Detection in High-Resolution Video Sensing

Sensors ◽  
2020 ◽  
Vol 20 (12) ◽  
pp. 3591 ◽  
Author(s):  
Haidi Zhu ◽  
Haoran Wei ◽  
Baoqing Li ◽  
Xiaobing Yuan ◽  
Nasser Kehtarnavaz
Keyword(s):  
High Resolution ◽  
Object Detection ◽  
Real Time ◽  
Moving Object Detection ◽  
Moving Object ◽  
Computationally Efficient ◽  
Real Time Processing ◽  
Time Processing ◽  
Video Frames ◽  
High Resolution Images

This paper addresses real-time moving object detection with high accuracy in high-resolution video frames. A previously developed framework for moving object detection is modified to enable real-time processing of high-resolution images. First, a computationally efficient method is employed, which detects moving regions on a resized image while maintaining moving regions on the original image with mapping coordinates. Second, a light backbone deep neural network in place of a more complex one is utilized. Third, the focal loss function is employed to alleviate the imbalance between positive and negative samples. The results of the extensive experimentations conducted indicate that the modified framework developed in this paper achieves a processing rate of 21 frames per second with 86.15% accuracy on the dataset SimitMovingDataset, which contains high-resolution images of the size 1920 × 1080.

scholarly journals Using a Balloon-Launched Unmanned Glider to Validate Real-Time WRF Modeling

Sensors ◽  
2019 ◽  
Vol 19 (8) ◽  
pp. 1914 ◽  
Author(s):  
Travis J. Schuyler ◽  
S. M. Iman Gohari ◽  
Gary Pundsack ◽  
Donald Berchoff ◽  
Marcelo I. Guzman
Keyword(s):  
High Resolution ◽  
Real Time ◽  
Gpu Computing ◽  
Atmospheric Temperature ◽  
Unmanned Aerial Systems ◽  
Processing Unit ◽  
Graphical Processing ◽  
Atmospheric Sampling ◽  
Wrf Modeling ◽  
Weather Models

The use of small unmanned aerial systems (sUAS) for meteorological measurements has expanded significantly in recent years. SUAS are efficient platforms for collecting data with high resolution in both space and time, providing opportunities for enhanced atmospheric sampling. Furthermore, advances in mesoscale weather research and forecasting (WRF) modeling and graphical processing unit (GPU) computing have enabled high resolution weather modeling. In this manuscript, a balloon-launched unmanned glider, complete with a suite of sensors to measure atmospheric temperature, pressure, and relative humidity, is deployed for validation of real-time weather models. This work demonstrates the usefulness of sUAS for validating and improving mesoscale, real-time weather models for advancements toward reliable weather forecasts to enable safe and predictable sUAS missions beyond visual line of sight (BVLOS).

scholarly journals A Flexible IoT Stream Processing Architecture Based on Microservices

Information ◽  
2020 ◽  
Vol 11 (12) ◽  
pp. 565
Author(s):  
Luca Bixio ◽  
Giorgio Delzanno ◽  
Stefano Rebora ◽  
Matteo Rulli
Keyword(s):  
Real Time ◽  
Query Language ◽  
Stream Processing ◽  
Heterogeneous Data ◽  
Core Level ◽  
Processing Unit ◽  
Reference Architecture ◽  
Real Time Processing ◽  
Stream Processing Engines ◽  
Processing Engine

The Internet of Things (IoT) has created new and challenging opportunities for data analytics. The IoT represents an infinitive source of massive and heterogeneous data, whose real-time processing is an increasingly important issue. IoT applications usually consist of multiple technological layers connecting ‘things’ to a remote cloud core. These layers are generally grouped into two macro levels: the edge level (consisting of the devices at the boundary of the network near the devices that produce the data) and the core level (consisting of the remote cloud components of the application). The aim of this work is to propose an adaptive microservices architecture for IoT platforms which provides real-time stream processing functionalities that can seamlessly both at the edge-level and cloud-level. More in detail, we introduce the notion of μ-service, a stream processing unit that can be indifferently allocated on the edge and core level, and a Reference Architecture that provides all necessary services (namely Proxy, Adapter and Data Processing μ-services) for dealing with real-time stream processing in a very flexible way. Furthermore, in order to abstract away from the underlying stream processing engine and IoT layers (edge/cloud), we propose: (1) a service definition language consisting of a configuration language based on JSON objects (interoperability), (2) a rule-based query language with basic filter operations that can be compiled to most of the existing stream processing engines (portability), and (3) a combinator language to build pipelines of filter definitions (compositionality). Although our proposal has been designed to extend the Senseioty platform, a proprietary IoT platform developed by FlairBit, it could be adapted to every platform based on similar technologies. As a proof of concept, we provide details of a preliminary prototype based on the Java OSGi framework.

Design Concept of Responsive Multithreaded Processor for Distributed Real-Time Control

2004 ◽  
Vol 16 (2) ◽  
pp. 194-199 ◽  
Author(s):  
Nobuyuki Yamasaki ◽  
Keyword(s):  
Real Time ◽  
Design Concept ◽  
Processing Unit ◽  
Home Automation ◽  
Real Time Control ◽  
Real Time Processing ◽  
Time Control ◽  
Multithreaded Processor ◽  
On Chip ◽  
Asic Chip

This paper describes the design concept of Responsive MultiThreaded (RMT) Processor for distributed real-time control that controls various embedded systems including robots, home automation, factory automation, etc. RMT processor integrates an 8-way multithreaded processor (RMT processing unit) for real-time processing, four sets of Responsive Link II for real-time communication, and I/O peripherals including DDR SDRAM I/Fs, DMAC, PCI64, USB2.0, IEEE1394, PWM generators, pulse counters, etc., into an ASIC chip. System designers can use various on-chip functions easily by connecting required I/Os to this chip directly. The designers can also realize distributed control systems by connecting several RMT processors with their own functions via Responsive Link II.

HIGH RESOLUTION, REAL-TIME LINE-FIELD FOURIER-DOMAIN INTERFEROMETRY

2012 ◽  
Vol 05 (02) ◽  
pp. 1250009
Author(s):  
QING XIAO ◽  
LING FU
Keyword(s):  
High Resolution ◽  
Real Time ◽  
Reconstruction Algorithm ◽  
Profile Measurement ◽  
Fourier Domain ◽  
Processing Unit ◽  
Line Field ◽  
Cross Sectional ◽  
Interferometric Method ◽  
The Cross

To increase the application potential in manufacturing process, such as monitoring the processing performance, the profile measurement should be provided in real-time display and with high resolution simultaneously. We propose a line-field Fourier-domain interferometric method (LFI), which combines the line-field microscope with spectral interferometer, for the surface cross-sectional profile measurement with no scan needed. The white light and objectives are employed to offer high axial and lateral resolution, respectively. In our system setup, the measurement could be implemented in real-time display of 10 frame/s, and the resolutions of the LFI system in X,Y, and Z directions are ~8 μm, ~3.2 μm, and ~1.4 μm, respectively. As a demonstration, the cross-sectional profiles of a microfluidic chip are tested. The graphics processing unit is also used to accelerate the reconstruction algorithm to achieve the real-time display of the cross-sectional profiles.

Sign in / Sign up

Export Citation Format

Share Document