(reading)Revisiting Visual Question Answering Baselines
来源:互联网 发布:产品经理数据分析手册 编辑:程序博客网 时间:2024/05/14 09:10
阅读后收获:对于VQA,使用MLP model,将I-Q-A作为输入做caption效果要好于以I-A作为输入,将I-A作为输入做caption效果要好于以Q-A作为输入,将Q-A作为输入做caption效果要好于以A作为输入,但是仅仅以A作为输入,在Visual7W telling task中就可以达到50.7%的准确性,说明了仅仅学习A分布的bias就可以取得不错的效果。另外结合不断填充扩大的Knowledge source,效果会进一步的提升,另外,通过iterative query去获取task-driven的信息,可以大大提高学习效果。
阅读全文
0 0
- (reading)Revisiting Visual Question Answering Baselines
- Paper Reading - Snap and ask: Answering Multimodal Question by Naming Visual Instance
- #Paper Reading# Summarizing Answers in Non-Factoid Community Question-Answering
- Hierarchical Question-Image Co-Attention for Visual Question Answering
- Hierarchical Question-Image Co-Attention for Visual Question Answering
- 视觉问答(Visual Question Answering)论文初步整理
- Dynamic Memory Networks for Visual and Textual Question Answering
- 论文笔记: Hierarchical Question-Image Co-Attention for Visual Question Answering
- Ask, Attend and Answer: Exploring Question-Guided Spatial Attention for Visual Question Answering
- READING NOTE: Wider or Deeper: Revisiting the ResNet Model for Visual Recognition
- 论文笔记 :Multimodal Compact Bilinear Pooling for Visual Question Answering and Visual Grounding
- 阅读笔记(Multimodal Compact Bilinear Pooling for Visual Question Answering and Visual Grounding)
- Multimodal Compact Bilinear Pooling for Visual Question Answering and Visual Grounding
- Ask Me Anything:Free-form Visual Question Answering Based on Knowledge from External Sources
- ABC-CNN: An Attention Based Convolutional Neural Network for Visual Question Answering
- 论文笔记:Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering
- hdu4096Universal Question Answering System(乱搞)
- Stacked Attention Networks for Image Question Answering
- ppt 2010 动画窗格 组合元素 使用方法
- register_chrdev_region函数解析
- n log n的最长上升子序列
- 抽象 关键字abstract与final使用
- android之频道管理
- (reading)Revisiting Visual Question Answering Baselines
- win10系统快捷键小结
- Java中的序列化Seriazable
- libCoAP的源码以及实现
- Struts基本配置文件
- Vue入门(一)-Vue实现选项卡效果
- 阿里笔试题
- 面试总结(1)
- ccf 201609-2 火车购票 (只有90分)