HSP-MFL: A High-level Semantic Property driven Multi-task Feature Learning Network for unsupervised person Re-ID

Unsupervised person re-identification aims to distinguish different pedestrians from discriminative representations on the basis of unlabeled data. Currently, most unsupervised Re-ID approaches explore visual representations to generate pseudo-labels for model’s training, which may suffer from background noise and semantic loss. To tackle this problem, this paper proposes a High-level Semantic Property driven Multi-task Feature Learning Network (HSP-MFL) to firstly introduce three high-level semantic properties for unsupervised person Re-ID. Technically, we design a novel Multiple Feature Fusion Module (MFFM) to deeply explore the complex correlation among multiple semantic and visual features to capture the discriminative feature cues, as well as a multi-task training scheme to generate robust fusion features. The architecture is quite simple and does not consume extra labeling costs. Extensive experiments on three datasets demonstrate that both high-level semantic properties and multi-task learning are effective in performance improvement, yielding SOTA mAPs for unsupervised person Re-ID.     

Variational Feature Representation-based Classification for face recognition with single sample per person

The single sample per person (SSPP) problem is of great importance for real-world face recognition systems. In SSPP scenario, there is always a large gap between a normal sample enrolled in the gallery set and the non-ideal probe sample. It is a crucial step for face recognition with SSPP to bridge the gap between the ideal and non-ideal samples. For this purpose, we propose a Variational Feature Representation-based Classification (VFRC) method, which employs the linear regression model to fit the variational information of a non-ideal probe sample with respect to an ideal gallery sample. Thus, a corresponding normal feature, which reserve the identity information of the probe sample, is obtained. A combination of the normal feature and the probe sample is used, which makes VFRC method more robust and effective for SSPP scenario. The experimental results show that VFRC method possesses higher recognition rate than other related face recognition methods.     

基于AVR单片机的语音识别系统设计

系统以AVR单片机为控制核心,实现对人的语音的识别控制。系统采用的主控芯片为Atmel公司的ATMEGA128,语音识别功能采用ICRoute公司的单芯片LD3320。LD3320内部集成语音识别算法,无需外部FLASH,RAM资源,可以很好地完成非特定人的语音识别任务。同时该芯片内部集成了MP3播放功能,支持MPEG等格式,可实现语音提示或MP3歌曲的播放功能。由于内部合有16位A／D、D／A转换器和功放电路,所以不需要外接功放电路就可以产生清晰的声音。该系统已经预留好各种接口,具有良好的扩展性。     

企业网络风险评估——基于人的因素研究

人类关于风险的研究古已有之。随着网络迅速渗透到社会各个领域,人们逐步认识到网络在推动社会发展的同时,也给社会各个领域带来诸多风险,主要表现在技术风险和人员管理的风险,要降低这些风险必须进行风险管理。论文通过分析,充分证实了基于人的网络风险评估方法对现有方法的有效扩展和补充,有效提高了网络风险评估的有效性和价值。     

Fine-grained-based multi-feature fusion for occluded person re-identification

Many previous occluded person re-identification(re-ID) methods try to use additional clues (pose estimation or semantic parsing models) to focus on non-occluded regions. However, these methods extremely rely on the performance of additional clues and often capture pedestrian features by designing complex modules. In this work, we propose a simple Fine-Grained Multi-Feature Fusion Network (FGMFN) to extract discriminative features, which is a dual-branch structure consisting of global feature branch and partial feature branch. Firstly, we utilize a chunking strategy to extract multi-granularity features to make the pedestrian information contained in it more comprehensive. Secondly, a spatial transformer network is introduced to localize the pedestrian’s upper body, and then introduce a relation-aware attention module to explore the fine-grained information. Finally, we fuse the features obtained from the two branches to obtain a more robust pedestrian representation. Extensive experiments verify the effectiveness of our method under the occlusion scenario.     

“本科+技师+工程师”人才培养模式的改革与实践

实践性教学是工科院校特别是应用型工科院校的重要教学环节,是培养学生创新意识、创新精神和创新能力的重要方法和手段。加强实践教学的改革与创新,需要构建独立的实践教学体系,整合实验教学内容、改革实践教学模式,理顺实践教学管理体制,建立科学合理的实践教学评估体系和质量保障体系。     

基于二阶混合注意力的行人重识别

基于注意力机制的行人重识别方法更多利用图像中 一阶信息,忽略了特征中二阶信息 ,不能挖掘特征图之间的相关性和细粒度信息。提出一种基于二阶混合注意力的行人重 识别算法(second-order mixed attention module,SMAN)。二阶混合注意力模块(second-order mixed attention module,SOMA)由二阶通道注意力(second-order channel attention,SOCA)和二阶空间注意力模 块(second-order information,SOSA)组成,该方法将全局协方差池函数嵌入到SOCA和SOSA模块中,学习特征中二阶信息 。SOCA模块学习特征图之间相关性,SOSA模块则重新为特征图分配权重,关注特征图空间域 的细粒度信息。SMAN算法在Market-1501和 DukeMTMC-ReID数据集上的首位准确率分别 为 94.3%和87.1%,mAP分别达到85.7%和74.5%,同时使用类激活图验证SOMA模块的影响 ,实验表 明SMAN算法充分利用特征图的通道域和空间域中二阶信息。算法的性能优于现有的一些基于 注意力机制行人重识别方法,甚至接近某些优秀的方法。     

Research on human blockage effect for indoor 26 GHz mm-wave communications

Based on 26 GHz indoor channel measurements, the blockage attenuation of human bodies was investigated. 2-ray and 4-ray human blockage models were proposed by calculation of the Vogler's multiple knife-edge diffraction field. Results show that 4-ray model fits quite well with measurement data. Human block attenuations are slightly bigger with 1 GHz than with 500 MHz bandwidth. In the blockage measurement of one person moving along the connec-tion line between the transmitter (TX) and receiver (RX), the smallest attenuation is found when the person is located at the middle of the connection line, and the biggest attenuation is observed when the person is nearest from the RX position. In the blockage measurements by multiple human bodies, the attenuation is bigger in the cases which the persons cross the TX-RX connection line with their front faces directed to the horn aperture at the RX than the cases with their lat-eral faces directed to the RX horn aperture. The blockage attenuation is larger with the increase of person numbers, about 5～8 dB more attenuation per person. The results can be used for design of mm-wave 26 GHz indoor communi-cations systems.     

广播电视编导专业人才培养的定位与思考

广播电视编导专业是高等教育中发展较为迅速的专业,如何健康有序的培养合格人才,也是专业学科建设中所面临的突出问题。结合专业特点、人才培养目标、培养模式等,文章从三个方面对该专业进行了分析、探讨,提出对目前高校广播电视编导专业方向建设认识与建议。     

基于GA的压缩域中全局运动估计及在字幕遮挡区域恢复中的应用

文章提出了一种直接使用压缩域中的运动矢量进行全局运动估计的方法,并用遗传算法优化输入运动矢量与全局运动参数所产生运动矢量的平均匹配误差.最终的实验结果表明本文的方法能够很好的估计出全局运动的参数.并且提出了一种运用全局/局部运动信息进行视频中字幕遮挡区域的恢复的方法.实验结果表明该方法取得了较好的视觉效果.