ID 原文 译文
40926 实验结果表明，该算法在峰值信噪比(PSNR)、结构相似度(SSIM)、特征相似度(FSIM)三种图像评价指标上均高于对比算法，取得了良好的去雾效果，有效地解决了去雾效果不理想的问题。 The experimental results show that the algorithm is superior to the comparison algorithm in three image evaluation indexes of peak signal-to-noise ratio(PSNR), structural similarity(SSIM) and feature similarity(FSIM), and it achieves a good defogging effect and effectively solves the problems of undesirable defogging effect
40927 设计了一种基于移动端深度学习目标检测算法MobileNetV2-SSDlite的小规模、高精度的人脸检测算法。 A small and high-precision face detection algorithm based on MobileNetV2-SSDlite is designed.
40928 在算法基础上，围绕硬件架构、片上缓存、性能及功耗等方面进行移动端专用型神经网络加速器的研究与设计; To improve the computing capability of mobile terminals, the study of the mobile algorithm specific neural network accelerator is carried out in terms of hardware architecture, on-chip memory, performance and power consumption.
40929 在ALINX AX7350 SOC平台上实现了实时多人脸检测加速器系统。 A real-time multi-face detection accelerator system is simulated and implemented on ALINX AX7350 SOC platform.
40930 结果表明，本系统在100 MHz时钟下，平均计算性能为56.01 GOPS，功耗为7.3 W。与现有的MobileNetV2-SSDlite加速器相比，运算速度提高了28.7%，资源平均节省了44.46%，功耗降低了26.3%。 The results show that under a 100 MHz clock, the face detection system has an average computing performance of 56.01 GOPS and a power consumption of 7.3 W. Compared with the existing MobileNetV2-SSDlite accelerator, the operation speed is improved by 28.7%, the resources are reduced by 44.46%, and the power consumption is reduced by 26.3%.
40931 在224×224分辨率下，达到了83.4 FPS，满足实时性需求。 The speed of the system in inferencing a 224×224 resolution picture reached 83.4 FPS, meet the demand of real-time performance.
40932 针对近眼显示器扩大视场角的需求，本文配合光学系统设计了一种基于FPGA的近眼显示实时预畸变系统。 To meet the requirement of enlarging the field of view angle of near-eye display, a real-time pre-distortion system for near-eye display based on FPGA is designed.
40933 本系统采用基于片外存储查找表结构的坐标校正方法，结合双线性插值的灰度校正算法，利用片内Block RAM资源设计了一种图像数据按奇偶行分别存储的数据缓存结构，减小了数据缓存量，降低了输出延时，实现了高分辨率、高帧率视频图像的实时预畸变处理。 This system adopts the coordinate correction method based on off-chip storage lookup table structure and combined with the bilinear interpolation algorithm for calculating grayscale correction, the use of chip designs an image data Block RAM resources stored separately according to the odd lines of data caching structure, reduced the data cache memory, reduces the output delay, has realized the high resolution, high frame rate of video image real-time pre distortion processing.
40934 实验结果表明，实时预畸变系统对光学系统产生的非线性几何畸变具有良好的校正效果，且处理延时小，显示图像清晰完整且可扩展性强。 The experimental results show that the real-time pre-distortion system has a good correction effect on the nonlinear geometric distortion generated by the optical system, and the processing delay is small, the display image is clear and complete, and the scalability is strong.
40935 本文提出了一种对云虚拟机进行异常检测的模型LsGrbmAd。 A model of Long Short Term Memory and Gaussian Boltzmann Machine Anomaly Detection(LsGrbmAd) is proposed in this paper.