《Table 2 Recognition accuracy w ith different sizes of convolution kernel》
本系列图表出处文件名:随高清版一同展现
《Individual Dairy Cattle Recognition Based on Deep Convolutional Neural Network》
To verify the effect of convolution kernel size tow ards recognition results,several different convolution kernel sizes are chosen for experiments referring to Le Net-5 model.When the convolution kernel takes 5×5,7×7,9×9,and 11×11 in turn,the results are show n in Table 2.When the size of convolution kernel is 5×5,the recognition accuracy is relatively low er.It is because the convolution kernel is too small and can’t cover effective feature extraction for dairy cattle individual.With increasing of the size of convolution kernel,the recognition accuracy show s a trend of rising and the kernel of 9×9 makes the recognition get a peak.Continuing to increase the size to 11×11,the recognition accuracy declines.The main reason is that the convolution kernel size is too big and some local features are lost.
图表编号 | XD0014946900 严禁用于非法目的 |
---|---|
绘制时间 | 2018.04.30 |
作者 | 张满囤、单新媛、于津苏、郭迎春、李睿文、徐明权 |
绘制单位 | School of Computer Science and Engineering,Hebei University of Technology、Hebei Province Key Laboratory of Big Data Calculation、School of Computer Science and Engineering,Hebei University of Technology、Hebei Province Key Laboratory of Big Data Calculation |
更多格式 | 高清、无水印(增值服务) |