Reading Notes Paper Reading [1] Satoshi Tsutsui, David Crandall(2017): Using Artificial Tokens to Control Languages for Multilingual Image Caption Generation. CoRR, abs/1706.06275, https://arxiv.org/abs/1706.06... Posted by Poole on June 28, 2017 Reading Notes Zhongliang Yang, Yu-Jin Zhang, Sadaqat ur Rehman & Yongfeng Huang(2017): Image Captioning with Object Detection and Localization. CoRR, abs/1706.02430, https://arxiv.org/abs/1706.02430 清华大学电子系... Posted by Poole on June 21, 2017 Some Notes 本笔记是对Google开源的Show-and-Tell代码的学习的一些笔记，并对TensorFlow的一些用法以及训练方法的整理。 输入 读取数据到数据队列当中。在输入数据很大的情况下，实际上只要构建出合理的输入队列，我们就可以不用去操心数据读入的事情。TensorFlow支持多种的数据读入方式，一种比较简单的方式是使用placeholder，但是这需要在使用的过程中feed_data... Posted by Poole on May 17, 2017 Documents for github repo. im2txtViaGAN github repo. link(https://github.com/mazm13/im2txtViaGAN) TODO LIST begin document figure out inputs ops target seqs’ information? difference with cross-entropy loss, think about how to use... Posted by Poole on May 11, 2017 Towards diverse and natural image descriptions via a conditional gan Posted by Poole on May 11, 2017 ← Newer Posts Older Posts →