系列文章目录

---

提示：更新中，一周左右更新完毕。需要具体课件的可私信

---

开幕式

主旨报告-1:大模型时代的机遇和挑战

沈向洋(香港科技大学)

主旨报告-2:以深度学习框架为牵引促进自主 AI生态发展

胡事民(清华大学)

主旨报告-3:从洞穴的影子到智能的光辉–连接和交互方式的改变塑造未来生活

大会特邀报告

1:图像生成和视频生成若干前沿技术探索

李鸿升(香港中文大学)

2:混合模型驱动的内容生成与具身智能

杨 易(浙江大学)

3:多模态视觉融合方法:是否存在性能极限

吴小俊(江南大学)

4:三维场景理解的前世、今生与未来

马月昕(上海科技大学)

年度进展评述

1.符号知识和大模型

陈华钓(浙江大学)

2.2智能体视觉

代季峰(清华大学)

---

提示：以下是本篇文章正文内容，下面案例可供参考

一、

二、第二天

1.引入库

2.读入数据

Tutorial 1:开放词汇视觉感知

讲者：李冠彬-中山大学
时间：第二天下午

一、课题背景介绍

---

---

![在这里插入图片描述](https://img-
blog.csdnimg.cn/direct/0fe51b99b65540e2b6282d2632b06171.png)

  开放词汇视觉感知 要求模型从海量图文对数据中学习视觉概念知识，并在实际开放 场景中实现不限类别 的视觉感知。

---

---

---

二、开放词汇分类

---

  利用海量图像-文本对，将图像与文本映射到同一嵌入空间，实现概念跨模态语义对齐。

特点：

图文关联弱，噪声较大
海量数据易获取
泛化能力强

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

三、开放词汇检测

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

四、开放词汇分割

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

五、下游任务应用

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

六、多模态大模型

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

---

七、总结与展望

---

总结

提示：这里对文章进行总结：

例如：以上就是今天要讲的内容，本文仅仅简单介绍了pandas的使用，而pandas提供了大量能使我们快速便捷地处理数据的函数和方法。