顾剑成的博客

Thinking will not overcome fear but action will.

MultiGPU-ComputeGradient-坑 Issue Environment: Tensorflow-1.2 问题是这样的，在使用多GPU进行数据并行的训练时，我们需要针对每个GPU设备单独分配部分的数据，我通过以下代码解决。 devices = [0,1,2] X = tf.placeholder(dtype=tf.float32,shape[None,224,22...

Posted by 顾剑成的博客 on January 10, 2018

论文阅读-Efficient Eye Typing with 9-direction Gaze Estimation

摘要

完整交互过程位置校准 Screen mode Screen size: 30x40 cm Distance: 25-40cm Cost Time: 10~20sec Off-screen mode Distance: 20~30cm ...

Posted by 顾剑成 on December 28, 2017

FEATURED TAGS

ABOUT ME

Experience the Life, Find Your Answer!

✉️ yzldw333@gmail.com