Chongyang Tao

Results 7 comments of Chongyang Tao

Do you mean the parameters in solver.prototxt or training network?

You can easily change [deploy.prototxt](https://github.com/chongyangtao/DeepHCCR/blob/master/googlenet_deploy.prototxt) to training network.

灰度图像的就可以用标准的矩阵填充或者是稀疏重建去做

内存我没太注意呢,不过应该不会超过128g呢,程序的数据读取使用的tfrecords,不会全部加载进去的