为探索大脑与视觉之间的联系,提高大脑活动重建视频的清晰度与准确性,提出了一种名为高质量脑电视频重建(high quality electroencephalogram video reconstruction,HQEEGVR)的方法进行脑电信号重建视频。首先,提出三分支脑电特征提取...为探索大脑与视觉之间的联系,提高大脑活动重建视频的清晰度与准确性,提出了一种名为高质量脑电视频重建(high quality electroencephalogram video reconstruction,HQEEGVR)的方法进行脑电信号重建视频。首先,提出三分支脑电特征提取网络——掩蔽时空频融合网络(masking spatio-temporal frequency fusion network,MSTFFNet)从脑电信号中提取大脑活动信息,深入挖掘大脑活动变化背后的语义,同时提取时空频信息;其次,引入跨模态对比学习,对齐脑电、文本、图像特征,以便生成阶段使用;然后,提出级联视频扩散模型,具体来说,先利用稳定扩散模型以脑电特征为条件生成参考视频帧,接着以视频帧为参考,融入运动矢量,引入视频扩散模型捕捉视频时间特征;最终生成高质量视频。结果表明,该模型在重建视频的主体、动作、颜色、语义等方面表现较好。可见利用脑电信号可以捕获大脑活动的视觉与语义信息,从而重建高保真度和视觉真实性的视频。展开更多
目的探究“非遗+AI”新模式下湖南土特产包装的创新设计形式,传承和发展非遗文化,提升土特产包装的文化内涵,提高传统包装的科技含量,以适应当代消费者不断提升的审美和消费需求。方法通过问卷调查、深度访谈等方法,分析“非遗+AI”模...目的探究“非遗+AI”新模式下湖南土特产包装的创新设计形式,传承和发展非遗文化,提升土特产包装的文化内涵,提高传统包装的科技含量,以适应当代消费者不断提升的审美和消费需求。方法通过问卷调查、深度访谈等方法,分析“非遗+AI”模式与当代湖南土特产包装的设计现状,了解消费者对包装设计的不同需求,总结设计策略,并基于“非遗+AI”新模式下机器学习和深度学习技术,借助SD(Stable Diffusion,稳定扩散模型)架构下的Lora模型(Low-RankAdaptation of Large Language Models,大语言模型的低秩适用方法),进行踏虎凿花在岳阳王鸽土特产包装设计中的创新应用实践。结果生成出紧跟新时代数字技术,驱动非遗文化活态传承的包装设计图案,从而实现湖南土特产包装的创新设计探索。结论基于“非遗+AI”模式下的SD模型技术,对泸溪踏虎凿花进行创新转化并应用在湖南土特产岳阳王鸽包装设计中,既活化了传统手工艺的表达形式,又创新了土特产的包装表现手法,为其他非遗文化元素和产品包装的有机融合提供了新路径。展开更多
Lithium metal stands out as an exceptionally promising anode material,boasting an extraordinarily high theoretical capacity and impressive energy density.Despite these advantageous characters,the issues of dendrite fo...Lithium metal stands out as an exceptionally promising anode material,boasting an extraordinarily high theoretical capacity and impressive energy density.Despite these advantageous characters,the issues of dendrite formation and volume expansion of lithium metal anodes lead to performance decay and safety concerns,significantly impeding their advancement towards widespread commercial viability.Herein,a lithium-rich Li-B-In composite anode with abundant lithophilic sites and outstanding structural stability is reported to address the mentioned challenges.The evenly distributed Li-In alloy in the bulk phase of anodes act as mixed ion/electron conductors and nucleation sites,facilitating accelerated Li ions transport dynamics and suppressing lithium dendrite formation.Additionally,these micron-sized Li-In particles in LiB fibers framework can enhance overall structural integrity and provide sufficient interior space to accommodate the volume changes during cycling.The electrochemical performance of Li-B-In composite anode exhibits long-term cyclability,superior rate performance and high-capacity retention.This work confirms that the synergy between a 3 D skeleton and hetero-metallic lithiophilic sites can achieve stable and durable lithium metal anodes,offering innovative insights for the practical deployment of lithium metal batteries.展开更多
文摘为探索大脑与视觉之间的联系,提高大脑活动重建视频的清晰度与准确性,提出了一种名为高质量脑电视频重建(high quality electroencephalogram video reconstruction,HQEEGVR)的方法进行脑电信号重建视频。首先,提出三分支脑电特征提取网络——掩蔽时空频融合网络(masking spatio-temporal frequency fusion network,MSTFFNet)从脑电信号中提取大脑活动信息,深入挖掘大脑活动变化背后的语义,同时提取时空频信息;其次,引入跨模态对比学习,对齐脑电、文本、图像特征,以便生成阶段使用;然后,提出级联视频扩散模型,具体来说,先利用稳定扩散模型以脑电特征为条件生成参考视频帧,接着以视频帧为参考,融入运动矢量,引入视频扩散模型捕捉视频时间特征;最终生成高质量视频。结果表明,该模型在重建视频的主体、动作、颜色、语义等方面表现较好。可见利用脑电信号可以捕获大脑活动的视觉与语义信息,从而重建高保真度和视觉真实性的视频。
文摘目的探究“非遗+AI”新模式下湖南土特产包装的创新设计形式,传承和发展非遗文化,提升土特产包装的文化内涵,提高传统包装的科技含量,以适应当代消费者不断提升的审美和消费需求。方法通过问卷调查、深度访谈等方法,分析“非遗+AI”模式与当代湖南土特产包装的设计现状,了解消费者对包装设计的不同需求,总结设计策略,并基于“非遗+AI”新模式下机器学习和深度学习技术,借助SD(Stable Diffusion,稳定扩散模型)架构下的Lora模型(Low-RankAdaptation of Large Language Models,大语言模型的低秩适用方法),进行踏虎凿花在岳阳王鸽土特产包装设计中的创新应用实践。结果生成出紧跟新时代数字技术,驱动非遗文化活态传承的包装设计图案,从而实现湖南土特产包装的创新设计探索。结论基于“非遗+AI”模式下的SD模型技术,对泸溪踏虎凿花进行创新转化并应用在湖南土特产岳阳王鸽包装设计中,既活化了传统手工艺的表达形式,又创新了土特产的包装表现手法,为其他非遗文化元素和产品包装的有机融合提供了新路径。
基金Project(2023YFC3905904)supported by the National Key Research and Development Program,ChinaProject(2220197000221)supported by the Team of Foshan National Hi-Tech Industrial Development Zone Industrialization Entrepreneurial Teams Program,ChinaProject(2024ZZTS0373)supported by the Central South University Graduate Student Autonomous Exploration Innovative Programme,China。
文摘Lithium metal stands out as an exceptionally promising anode material,boasting an extraordinarily high theoretical capacity and impressive energy density.Despite these advantageous characters,the issues of dendrite formation and volume expansion of lithium metal anodes lead to performance decay and safety concerns,significantly impeding their advancement towards widespread commercial viability.Herein,a lithium-rich Li-B-In composite anode with abundant lithophilic sites and outstanding structural stability is reported to address the mentioned challenges.The evenly distributed Li-In alloy in the bulk phase of anodes act as mixed ion/electron conductors and nucleation sites,facilitating accelerated Li ions transport dynamics and suppressing lithium dendrite formation.Additionally,these micron-sized Li-In particles in LiB fibers framework can enhance overall structural integrity and provide sufficient interior space to accommodate the volume changes during cycling.The electrochemical performance of Li-B-In composite anode exhibits long-term cyclability,superior rate performance and high-capacity retention.This work confirms that the synergy between a 3 D skeleton and hetero-metallic lithiophilic sites can achieve stable and durable lithium metal anodes,offering innovative insights for the practical deployment of lithium metal batteries.
