scholarly journals A Real-Time FPGA Accelerator Based on Winograd Algorithm for Underwater Object Detection

Electronics ◽  
2021 ◽  
Vol 10 (23) ◽  
pp. 2889
Author(s):  
Liangwei Cai ◽  
Ceng Wang ◽  
Yuan Xu
Keyword(s):  
Energy Efficiency ◽  
Object Detection ◽  
Real Time ◽  
Autonomous Underwater Vehicles ◽  
Light Attenuation ◽  
Underwater Vehicles ◽  
Fpga Implementation ◽  
Underwater Imaging ◽  
Feature Extractor ◽  
Underwater Object

Real-time object detection is a challenging but crucial task for autonomous underwater vehicles because of the complex underwater imaging environment. Resulted by suspended particles scattering and wavelength-dependent light attenuation, underwater images are always hazy and color-distorted. To overcome the difficulties caused by these problems to underwater object detection, an end-to-end CNN network combined U-Net and MobileNetV3-SSDLite is proposed. Furthermore, the FPGA implementation of various convolution in the proposed network is optimized based on the Winograd algorithm. An efficient upsampling engine is presented, and the FPGA implementation of squeeze-and-excitation module in MobileNetV3 is optimized. The accelerator is implemented on a Zynq XC7Z045 device running at 150 MHz and achieves 23.68 frames per second (fps) and 33.14 fps when using MobileNetV3-Large and MobileNetV3-Small as the feature extractor. Compared to CPU, our accelerator achieves 7.5×–8.7× speedup and 52×–60× energy efficiency.

Laser-based imaging applications in turbid waters

2019 ◽  
Vol 8 (6) ◽  
pp. 461-467
Author(s):  
Thomas Scholz ◽  
Martin Laurenzis ◽  
Frank Christnacher
Keyword(s):  
Laser Scanning ◽  
Autonomous Underwater Vehicles ◽  
Underwater Vehicles ◽  
Imaging Systems ◽  
Underwater Imaging ◽  
The Past ◽  
Turbid Waters ◽  
Active Imaging ◽  
Underwater Laser ◽  
Processing Techniques

Abstract Underwater laser-based imaging systems and data-processing techniques matured during the past decade. Active imaging systems can, nowadays, be integrated into platforms like remote-operated vehicles (ROV) or autonomous underwater vehicles (AUV). This article gives an overview of different civil and naval applications in underwater imaging with respect to underwater laser scanning (ULS) and laser gated viewing (LGV). Special emphasis has to be given to the environmental conditions, for example, the influence of the local and seasonal dependence of the turbidity with regard to the optical underwater channel. On the basis of tank and sea experiments, advanced techniques for 3D laser oblique scanning (LOS) and possibilities of contrast enhancements for gated viewing are presented.

Real-Time Underwater Object Detection Based on DC Resistivity Method

2016 ◽  
Vol 54 (11) ◽  
pp. 6833-6842 ◽  
Author(s):  
Sung-Ho Cho ◽  
Hyun-Key Jung ◽  
Hyosun Lee ◽  
Hyoungrea Rim ◽  
Seong Kon Lee
Keyword(s):  
Object Detection ◽  
Real Time ◽  
Dc Resistivity ◽  
Resistivity Method ◽  
Underwater Object

scholarly journals Design, Analysis and Development of Autonomous Underwater vehicle

2019 ◽  
Vol 9 (2) ◽  
pp. 3805-3810
Keyword(s):  
Object Recognition ◽  
Motion Control ◽  
Water Surface ◽  
Autonomous Underwater Vehicle ◽  
Autonomous Underwater Vehicles ◽  
Underwater Vehicle ◽  
Underwater Vehicles ◽  
Heading Control ◽  
Underwater Object ◽  
Institute Of Technology

Autonomous Underwater Vehicles (AUV) are slowly operated unmanned robots which Capable of propelling on pre-defined mission tracks independently under the water surface and are frequently used for oceanographic exploration, bathymetric surveys and defense applications. This AUV can perform underwater object recognition and obstacle avoidance with the use of appropriate sensors and devices. Vidyut is a miniature AUV developed at Sri Sairam Institute of Technology. The vehicle is equipped with six thrusters which allow for motion control in 6 Dof and has a non-conventional single hull heavy bottom hydrodynamic design. This paper discusses different aspects of the vehicle's unique design. The output of the Arduino Uno controller has been discussed for continuous depth and heading control.

scholarly journals An edge detection mechanism using L*A*B color-based contrast enhancement for underwater images

2020 ◽  
Vol 18 (1) ◽  
pp. 41
Author(s):  
M Sudhakara ◽  
M Janaki Meena
Keyword(s):  
Edge Detection ◽  
Contrast Enhancement ◽  
Color Change ◽  
Autonomous Underwater Vehicles ◽  
Underwater Vehicles ◽  
Detection Algorithm ◽  
Input Image ◽  
Underwater Imaging ◽  
Detection Mechanism ◽  
Edge Detection Algorithm

In Ocean investigations, particularly those deployed by the Autonomous Underwater Vehicles, underwater object detection and recognition is an essential task. Edge detection places a key role and considered one of the pre-processing techniques for several deep learning applications. In an underwater environment, the illumination of light, turbulence in the water, suspended particles present in the seafloor are challenging issues to acquire the quality image. The two major problems in underwater imaging are light scattering and color change. In the former case, the vision sensors connected to the underwater vehicles or dive lights used by the divers themselves cause light dispersion and shadows in the seafloor. In the latter case, the occurrence of color distortion is mainly due to the attenuation of the light, hence the images are having dominant colors in the latter case. The conventional techniques are failed to detect the quality edges in the case of underwater images. Our mechanism focused, instead of applying the edge detection algorithm on the input image directly, it is better to apply edge detection algorithm after color correction and contrast enhancement using L*A*B model. Qualitative and quantitative test results demonstrate that the proposed mechanism is giving better results compared with state-of-the-art methods.

Sign in / Sign up

Export Citation Format

Share Document