1.卷积核 卷积核有三个尺寸长度、宽度、深度。其中深度一般默认为channel_in的大小。 2.channel_out 卷积核个数就等于channel_out个数。 参考blog:如何理解卷积神经网络中的通道(channel)_神经网络通道数-CSDN博客https://blog.csdn.net/weixin_38481963/article/details/109920512?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522170979253416800215029586%2522%252C%2522scm%2522%253A%252220140713.130102334..%2522%257D&request_id=170979253416800215029586&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~all~sobaiduend~default-3-109920512-null-null.142^v99^pc_search_result_base9&utm_term=%E5%8D%B7%E7%A7%AF%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%20channel&spm=1018.2226.3001.4187