摘要 :
This paper reviews the NTIRE 2022 challenge on efficient single image super-resolution with focus on the proposed solutions and results. The task of the challenge was to super-resolve an input image with a magnification factor of ...
展开
This paper reviews the NTIRE 2022 challenge on efficient single image super-resolution with focus on the proposed solutions and results. The task of the challenge was to super-resolve an input image with a magnification factor of ×4 based on pairs of low and corresponding high resolution images. The aim was to design a network for single image super-resolution that achieved improvement of efficiency measured according to several metrics including runtime, parameters, FLOPs, activations, and memory consumption while at least maintaining the PSNR of 29.00dB on DIV2K validation set. IMDN is set as the baseline for efficiency measurement. The challenge had 3 tracks including the main track (runtime), sub-track one (model complexity), and sub-track two (overall performance). In the main track, the practical runtime performance of the submissions was evaluated. The rank of the teams were determined directly by the absolute value of the average runtime on the validation set and test set. In sub-track one, the number of parameters and FLOPs were considered. And the individual rankings of the two metrics were summed up to determine a final ranking in this track. In sub-track two, all of the five metrics mentioned in the description of the challenge including runtime, parameter count, FLOPs, activations, and memory consumption were considered. Similar to sub-track one, the rankings of five metrics were summed up to determine a final ranking. The challenge had 303 registered participants, and 43 teams made valid submissions. They gauge the state-of-the-art in efficient single image super-resolution.
收起
摘要 :
The opposed piston opposed cylinder (opoc engine concept has been demonstrated as an engine concept with high specific power density and high-power-to-volume ratio. The engine has several potential applications, including use as a...
展开
The opposed piston opposed cylinder (opoc engine concept has been demonstrated as an engine concept with high specific power density and high-power-to-volume ratio. The engine has several potential applications, including use as an auxiliary power unit (APU) in various commercial and military applications and as the primary power source for small unmanned air vehicles (UAVs). An engine in this power range operating on heavy fuels (e.g., JP5, JP8, DF2) is not typically available. The engine uses a two-cycle supercharged uniflow scavenging system with asymmetric port timing and will run at speeds between 8,000 and 12,000 rpm. The unique design of the opoc engine produces a piston speed that is half the speed of a typical crankshaft engine running at the same speed. Uniflow scavenging produces gas exchange efficiencies rivaling those of four-cycle engines. The design also leads to reduced in-cylinder heat losses. Furthermore, the opoc engine is fully balanced. These significant benefits lead to an extremely lightweight and compact engine design. Extensive 1-D gas exchange and 3-D scavenge process simulation and optimizations were used to design the combustion system. Traditional Computer-Aided Engineering (CAE) practices in the area of Multi-Body Systems (MBS) simulation and thermo-mechanical Finite Element Analysis (FEA) were adapted to support the design and development of this engine. This paper provides an overview of the design process with focus on the unique features of the opoc engine and presents the system engineering techniques used in its design along with some results.
收起
摘要 :
The power system is becoming larger and larger, and the operation is more frequent, which puts higher requirements on the real-time performance of analysis and calculation. The graph database is a new type of database that origina...
展开
The power system is becoming larger and larger, and the operation is more frequent, which puts higher requirements on the real-time performance of analysis and calculation. The graph database is a new type of database that originated from the parallel processing of massive data in the Internet. The data model can visually express the topology of the grid and easily realize parallel traversal query. Firstly, the characteristics of graph database are introduced from the aspects of data model and data query, and the potential advantages of applying it to large-scale power system analysis and calculation are analyzed. Secondly, the design method of power system modeling is presented for satisfying the guidelines of integrity, consistency and efficiency conforming with CIM/E standard. Finally, a parallel power network topology analysis algorithm is implemented based on the graph database model for a provincial grid. The calculation results of the actual large-scale provincial power grid show that the proposed method can significantly improve the topology search efficiency.
收起
摘要 :
The power system is becoming larger and larger, and the operation is more frequent, which puts higher requirements on the real-time performance of analysis and calculation. The graph database is a new type of database that origina...
展开
The power system is becoming larger and larger, and the operation is more frequent, which puts higher requirements on the real-time performance of analysis and calculation. The graph database is a new type of database that originated from the parallel processing of massive data in the Internet. The data model can visually express the topology of the grid and easily realize parallel traversal query. Firstly, the characteristics of graph database are introduced from the aspects of data model and data query, and the potential advantages of applying it to large-scale power system analysis and calculation are analyzed. Secondly, the design method of power system modeling is presented for satisfying the guidelines of integrity, consistency and efficiency conforming with CIM/E standard. Finally, a parallel power network topology analysis algorithm is implemented based on the graph database model for a provincial grid. The calculation results of the actual large-scale provincial power grid show that the proposed method can significantly improve the topology search efficiency.
收起
摘要 :
Hyperspectral image (HSI) recovery from a single RGB image has attracted much attention, whose performance has recently been shown to be sensitive to the camera spectral sensitivity (CSS). In this paper, we present an efficient co...
展开
Hyperspectral image (HSI) recovery from a single RGB image has attracted much attention, whose performance has recently been shown to be sensitive to the camera spectral sensitivity (CSS). In this paper, we present an efficient convolutional neural network (CNN) based method, which can jointly select the optimal CSS from a candidate dataset and learn a mapping to recover HSI from a single RGB image captured with this algorithmically selected camera. Given a specific CSS, we first present a HSI recovery network, which accounts for the underlying characteristics of the HSI, including spectral nonlinear mapping and spatial similarity. Later, we append a CSS selection layer onto the recovery network, and the optimal CSS can thus be automatically determined from the network weights under the nonnegative sparse constraint. Experimental results show that our HSI recovery network outperforms state-of-the-art methods in terms of both quantitative metrics and perceptive quality, and the selection layer always returns a CSS consistent to the best one determined by exhaustive search.
收起
摘要 :
Hyperspectral image (HSI) recovery from a single RGB image has attracted much attention, whose performance has recently been shown to be sensitive to the camera spectral sensitivity (CSS). In this paper, we present an efficient co...
展开
Hyperspectral image (HSI) recovery from a single RGB image has attracted much attention, whose performance has recently been shown to be sensitive to the camera spectral sensitivity (CSS). In this paper, we present an efficient convolutional neural network (CNN) based method, which can jointly select the optimal CSS from a candidate dataset and learn a mapping to recover HSI from a single RGB image captured with this algorithmically selected camera. Given a specific CSS, we first present a HSI recovery network, which accounts for the underlying characteristics of the HSI, including spectral nonlinear mapping and spatial similarity. Later, we append a CSS selection layer onto the recovery network, and the optimal CSS can thus be automatically determined from the network weights under the nonnegative sparse constraint. Experimental results show that our HSI recovery network outperforms state-of-the-art methods in terms of both quantitative metrics and perceptive quality, and the selection layer always returns a CSS consistent to the best one determined by exhaustive search.
收起
摘要 :
Most existing cloud workflow modeling approaches focus on how to support business process customization for multiple tenants. But there are many factors to be considered in actual cloud workflow modeling, such as how to model inte...
展开
Most existing cloud workflow modeling approaches focus on how to support business process customization for multiple tenants. But there are many factors to be considered in actual cloud workflow modeling, such as how to model interaction between different tenants' business processes while protecting their privacy respectively, and how to facilitate the reuse of common process fragments for different stakeholder roles. To address these issues, this paper proposes a cloud workflow modeling framework using extended proclets. The framework provides a hierarchical management mechanism to isolate the private data of each tenant's business process and adopts two-level-channel transmission protocol to explicitly model interaction between different tenants' business processes. At last, a case study is carried out to illustrate its modeling capability and feasibility.
收起
摘要 :
Most existing cloud workflow modeling approaches focus on how to support business process customization for multiple tenants. But there are many factors to be considered in actual cloud workflow modeling, such as how to model inte...
展开
Most existing cloud workflow modeling approaches focus on how to support business process customization for multiple tenants. But there are many factors to be considered in actual cloud workflow modeling, such as how to model interaction between different tenants' business processes while protecting their privacy respectively, and how to facilitate the reuse of common process fragments for different stakeholder roles. To address these issues, this paper proposes a cloud workflow modeling framework using extended proclets. The framework provides a hierarchical management mechanism to isolate the private data of each tenant's business process and adopts two-level-channel transmission protocol to explicitly model interaction between different tenants' business processes. At last, a case study is carried out to illustrate its modeling capability and feasibility.
收起
摘要 :
This paper presents the Communication-Avoiding 3D Matrix Multiplication (CA3DMM) algorithm, a simple and novel algorithm that has optimal or near-optimal communication cost. CA3DMM is based on a unified view of parallel matrix mul...
展开
This paper presents the Communication-Avoiding 3D Matrix Multiplication (CA3DMM) algorithm, a simple and novel algorithm that has optimal or near-optimal communication cost. CA3DMM is based on a unified view of parallel matrix multiplication. Such a view generalizes 1D, 2D, and 3D matrix multiplication algorithms to reduce the data exchange volume for different shapes of input matrices. CA3DMM further minimizes the actual communication costs by carefully organizing its communication patterns. CA3DMM is much simpler than some other generalized 3D algorithms, and CA3DMM does not require low-level optimization. Numerical experiments show that CA3DMM has good parallel scalability and has similar or better performance when compared to state-of-the-art PGEMM implementations for a wide range of matrix dimensions and number of processes.
收起
摘要 :
Focusing on the requirements of amphibious operations, this paper mainly analyzes the composition of unmanned force, typical combat missions, command and control relations, etc. in amphibious operations. Combined with the analysis...
展开
Focusing on the requirements of amphibious operations, this paper mainly analyzes the composition of unmanned force, typical combat missions, command and control relations, etc. in amphibious operations. Combined with the analysis of the development trend of unmanned equipment and command and control systems, this paper constructs the architecture of intelligent cooperative command and control system of unmanned equipment in amphibious operations, and proposes the system composition and the key technologies to break through, including mission planning, route planning, multi-source information intelligence processing, decision, integrated control of large-scale cross domain, effect evaluation of land-attacking based on image understanding. It can provide reference for the development and standardization of intelligent cooperative command and control system of unmanned equipment.
收起