基于SSU-SGD的动态手持物体识别

连续目标中包含更加丰富的信息，为了更好地获取动态手持物体中的视觉信息，以不同背景下的动态手持物体为目标，基于步长自学习更新的SGD算法（简称SSU-SGD）提出了适用于动态手持物体识别的三个基准，通过自学习出不同的步长，分别在已知类、未知类和已知对象的基础上进行巩固训练，用于后续的动态手持物体识别中。用AlexNet和VGG网络对三个不同基准下的naive策略和累积策略进行了编程实验与仿真，经实验验证，该方法可以有效提高运行速度和训练精确度，并且有效提高了动态手持物体识别过程的实时性，可以进一步应用于实际。

Dynamic hand-held object recognition based on SSU-SGD

Continuous object contains richer information. In order to obtain more meaningful visual information in dynamic hand-held objects, targeting dynamic hand-held objects in different backgrounds, based on the step-size self-learning update SGD(SSU-SGD), this paper proposed three benchmarks for dynamic hand-held object recognition. Through self-learning of different step sizes, it consolidated the train on the basis of known classes, unknown classes, and known objects for subsequent dynamic hand-held object recognition. Using AlexNet and VGG networks, it performed the programming experiments and simu-lations of naive and cumulative strategies under the three different benchmarks. After experimental verification, this method can effectively improve the running speed and training accuracy and effectively improve the real-time nature of the dynamic hand-held object recognition process, it can be further applied to the actual.