Quality assessment for color correction-based stitched images via bi-directional matching

With the deepening of social information, the panoramic image has drawn a significant interest of viewers and researchers as it can provide a very wide field of view (FoV). Since panoramic images are usually obtained by capturing images with the overlapping regions and then stitching them together, image stitching plays an important role in generating panoramic images. In order to effectively evaluate the quality of stitched images, a novel quality assessment method based on bi-directional matching is proposed for stitched images. Specifically, dense correspondences between the testing and benchmark stitched images are first established by bi-directional SIFT-flow matching. Then, color-aware, geometric-aware and structure-aware features are respectively extracted and fused via support vector regression (SVR) to obtain the final quality score. Experiments on our newly constructed database and ISIQA database demonstrate that the proposed method can achieve comparable performance compared with the conventional blind quality metrics and the quality metrics specially designed for stitched images.     

2种侧围加强内框总成检具结构对比分析

概述了侧围加强内框总成与周边零件的匹配关系,从检具操作的人机工程出发,描述了侧围加强内框总成检具的2种定位方式和结构设计,对比了2种定位方式对于检测结果的影响,总结了其优点和弊端,结果表明侧围加强内框总成检具采用车内方向朝下的定位方式效果更好。     

结合预训练模型和语言知识库的文本匹配方法

针对文本匹配任务,该文提出一种大规模预训练模型融合外部语言知识库的方法。该方法在大规模预训练模型的基础上,通过生成基于WordNet的同义—反义词汇知识学习任务和词组—搭配知识学习任务引入外部语言学知识。进而,与MT-DNN多任务学习模型进行联合训练,以进一步提高模型性能。最后利用文本匹配标注数据进行微调。在MRPC和QQP两个公开数据集的实验结果显示,该方法可以在大规模预训练模型和微调的框架基础上,通过引入外部语言知识进行联合训练有效提升文本匹配性能。     

面向智能客服系统的情感分析技术

该文以阿里小蜜为例,对智能客服系统中的情感分析技术进行比较全面的介绍,包括情感分析算法模型的原理及其在智能客服系统的多个应用场景中的实际落地使用方式和效果分析。智能客服在解决客户高频业务问题的同时,也需要给客户提供多维度的、具有类人能力的助理、导购、语聊和娱乐等服务,提高客户对智能客服机器人的整体满意度。在此过程中,情感分析技术在机器人类人能力建设中起到了至关重要的作用。该文围绕智能客服系统中人机结合的服务形式,从六个维度总结和介绍了情感分析技术在智能客服系统中的应用场景,包括用户情感检测、用户情感安抚、情感生成式语聊、客服服务质检、会话满意度预估和智能人工入口。     

诱发IAF夹杂物与原晶粒的尺寸匹配关系

 对于氧化物冶金船板钢而言，细小弥散分布的夹杂物会对其原始晶粒产生钉扎和诱发晶内铁素体的双重作用，且研究发现诱发晶内铁素体的夹杂物尺寸与原始奥氏体晶粒尺寸存在一定的匹配关系。围绕上述问题进行了系统的分析研究和数据统计,结果表明，不同晶粒尺寸下，诱发晶内铁素体的夹杂物尺寸与原奥氏体晶粒度间存在近似线性关系，且随着奥氏体晶粒度的增大，诱发晶内铁素体的夹杂物尺寸则相应减小。     

A comparison of visual ability and its importance awareness between novice and experienced drivers

The visual ability factors of peripheral perception, visual reaction and visual pursuit ability are important for safe driving. Thus, this study explored the influence of driving experience on these visual ability factors. Sixty-five drivers, including novice and experienced drivers, participated in this study. The visual ability levels of the three factors were measured by the Vienna Test System. In addition, driver's awareness of the three factors' importance for safe driving was measured by questionnaire items. Results showed that out of the three factors, drivers' peripheral perception ability was correlated significantly with their driving experience. Experienced drivers had higher peripheral perception ability than that of novice drivers. In addition, compared with novice drivers, experienced drivers believed that peripheral perception was more important whereas visual reaction was less important. It is suggested that higher attention should be paid to peripheral perception in driving training and novice drivers should increase their importance awareness for this factor.     

用于快速服装搭配的FMatchNet算法

目的 针对现有服装搭配系统中，提取服装图像深度特征进行搭配所需时间过长的问题，提出了一种新的FMatchNet网络提取哈希特征进行服装快速搭配的方法。方法 首先采用快速区域卷积神经网络（Faster-RCNN）方法检测出图像中的服装，用此服装进行搭配可以最大限度地保留服装信息并消除背景信息的干扰。然后用深度卷积神经网络提取服装的深度特征并产生服装的哈希码，采用查询扩展的方法完成服装搭配。模型采用Siamese网络的训练方法使哈希码尽可能保留服装图像的语义信息。另外，由于目前国际上缺少大型时尚服装数据库，本文扩建了一个细粒度标注的时尚服装数据库。结果 在FClothes数据库上验证本文方法并与目前流行的方法进行对比，本文方法在哈希长度为16时，上、下服装搭配方面的准确度达到了50.81%，搭配速度相对于基本准线算法提高了近3倍。结论 针对大规模服装搭配问题，提出一种新的FMatchNet网络提取特征进行服装快速搭配的方法，提高了服装搭配的精度和速度，适用于日常服装搭配。     

基于改进网格运动统计特征的图像匹配算法

为了解决尺度不变特征变换（SIFT）算法在图像匹配中匹配正确率低、耗时长等问题，提出一种基于改进网格运动统计特征RANSAC-GMS的图像匹配算法。首先，利用快速旋转不变性特征（ORB）算法对图像进行预匹配，对预匹配的特征点采用网格运动统计（GMS）来支持估计量以实现正确匹配点与错误匹配点的区分；然后，采用改进的随机抽样一致性（RANSAC）算法通过匹配点间的距离相似性对特征点进行筛选，并采用评价函数对筛选后的新数据集进行重新整理，进而实现对误匹配点的剔除。采用Oxford标准图库和现实中拍摄的图像对图像匹配算法进行测试对比，实验结果表明，所提算法在图像匹配中的平均匹配正确率达到91%以上；与GMS、SIFT、ORB等算法相比，该改进算法的近景匹配正确率和远景匹配正确率分别最少提高了16.15个百分点和3.56个百分点，说明它能有效剔除误匹配点，进一步提高图像匹配精度。