A new protein wood adhesive was studied with Camellia oleifera protein.Formaldehyde and N-(2)-L-Ala-L-Gln(LAG)were used as the model compounds of amino resins and Camellia oleifera protein,aiming to provide scientific...A new protein wood adhesive was studied with Camellia oleifera protein.Formaldehyde and N-(2)-L-Ala-L-Gln(LAG)were used as the model compounds of amino resins and Camellia oleifera protein,aiming to provide scientific foundations for the improvement and applications of Camellia oleifera protein adhesive by the reaction of model compounds.The experimental results demonstrate that,under alkaline conditions,formaldehyde is easier to react with Camellia oleifera protein by quicker reaction and lower curing temperature.Under acid conditions,amino hydroxymethylated structure of aliphatic series from LAG is difficult to form stable reactive intermediates and further polycondensation.Hydroxymethylation of end acylamino and peptide bond amino from LAG is relatively weak.Under alkaline conditions,the free aliphatic amino and acylamino of LAG both can make hydroxymethylation reaction with formaldehyde.In the polycondensation,hydroxymethyl amide is the initial structure and the reactive intermediate is produced by E1cb reaction of hydroxymethyl amide.Methylene bridge bonds and methylene ether bonds are structures of the polycondensation products,which are competing reactions.The former is mainly formed by the reaction between alkaline reactive intermediate and amino of aliphatic series,and the latter is produced by the reaction of reactive intermediate and amino of hydroxymethyl aliphatic series with hydroxymethyl amide.展开更多
文摘数据增广是提升深度学习模型性能的有效方法之一。针对多类别目标检测任务中检测性能不平衡问题,提出一种针对“短板类别”(检测性能远低于模型平均检测性能的类别)的离线数据增广方法。受Cannikin’s Law的启发,采用基于复制粘贴(copy-paste)机制的场景多样性增广方法。随机采集训练集中“短板类别”实例区域,通过相似性度量机制选取训练集中增广目标样本进行随机粘贴。为了降低随机粘贴导致的遮挡问题,采用基于自遮挡(cut-replace)机制的增广方法提升模型遮挡表达能力。通过截取样本自身区域,对特征表达最显著区域进行遮挡。实验表明,FCOS目标检测框架在PASCAL VOC数据上的平均检测精度(mean average precision,mAP)从79.10%提升到83.90%,其中短板类别更为显著,提升了20.8个百分点。在MS-COCO数据上平均检测精度提升了0.9个百分点。
基金Funded by the National Natural Science Foundation of China (32160348)the Department Program of Guizhou Province (ZK[2021]162 and[2019]2325)+1 种基金the Special Project of“Doctor Professor Service Group of Kaili University (BJFWT201906)the Cultivation Project of Guizhou University of China ([2019]37)。
文摘A new protein wood adhesive was studied with Camellia oleifera protein.Formaldehyde and N-(2)-L-Ala-L-Gln(LAG)were used as the model compounds of amino resins and Camellia oleifera protein,aiming to provide scientific foundations for the improvement and applications of Camellia oleifera protein adhesive by the reaction of model compounds.The experimental results demonstrate that,under alkaline conditions,formaldehyde is easier to react with Camellia oleifera protein by quicker reaction and lower curing temperature.Under acid conditions,amino hydroxymethylated structure of aliphatic series from LAG is difficult to form stable reactive intermediates and further polycondensation.Hydroxymethylation of end acylamino and peptide bond amino from LAG is relatively weak.Under alkaline conditions,the free aliphatic amino and acylamino of LAG both can make hydroxymethylation reaction with formaldehyde.In the polycondensation,hydroxymethyl amide is the initial structure and the reactive intermediate is produced by E1cb reaction of hydroxymethyl amide.Methylene bridge bonds and methylene ether bonds are structures of the polycondensation products,which are competing reactions.The former is mainly formed by the reaction between alkaline reactive intermediate and amino of aliphatic series,and the latter is produced by the reaction of reactive intermediate and amino of hydroxymethyl aliphatic series with hydroxymethyl amide.
