felixchenfy / Realtime-Action-Recognition

Apply ML to the skeletons from OpenPose; 9 actions; multiple people. (WARNING: I'm sorry that this is only good for course demo, not for real world applications !!! Those ary very difficult !!!)
MIT License
875 stars 256 forks source link

您好!可以请教一下是否有相关文章或者资料对这个项目的介绍分析吗 #82

Open QueenW1998 opened 2 years ago

felixchenfy commented 2 years ago

hi,抱歉除了README的介绍以外就没有啦。 我当时没有看什么文献 --- 这个项目只是应用了一下机器学习来做分类

QueenW1998 commented 2 years ago

感谢回复,我目前在做这方面的研究,有想要继续请教的地方,如果可以的话能否添加一个联系方式?qq_1543748379

felixchenfy commented 2 years ago

不了吧😂 我自毕业以来就没做过机器学习了,水平不行。如果真要探讨,我们邮件讨论吧,felixchenfy@gmail.com,谢谢

QueenW1998 commented 2 years ago

好的,师兄。我现在正在做openpose人体姿态动作识别研究,您能大概给我讲一下这个项目的整体流程和运用到的一些算法吗。谢谢师兄!

------------------ 原始邮件 ------------------ 发件人: "felixchenfy/Realtime-Action-Recognition" @.>; 发送时间: 2022年5月14日(星期六) 下午4:00 @.>; @.**@.>; 主题: Re: [felixchenfy/Realtime-Action-Recognition] 您好!可以请教一下是否有相关文章或者资料对这个项目的介绍分析吗 (Issue #82)

不了吧😂 @.***,谢谢

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

felixchenfy commented 2 years ago

流程/算法我写在README里了啊: https://github.com/felixchenfy/Realtime-Action-Recognition#1-algorithm https://github.com/felixchenfy/Data-Storage/blob/master/EECS-433-Pattern-Recognition/FeiyuChen_Report_EECS433.pdf

QueenW1998 commented 2 years ago

好的,谢谢师兄,我先学习一下。

------------------ 原始邮件 ------------------ 发件人: "felixchenfy/Realtime-Action-Recognition" @.>; 发送时间: 2022年5月15日(星期天) 凌晨2:09 @.>; @.**@.>; 主题: Re: [felixchenfy/Realtime-Action-Recognition] 您好!可以请教一下是否有相关文章或者资料对这个项目的介绍分析吗 (Issue #82)

流程/算法我写在README里了啊: https://github.com/felixchenfy/Realtime-Action-Recognition#1-algorithm https://github.com/felixchenfy/Data-Storage/blob/master/EECS-433-Pattern-Recognition/FeiyuChen_Report_EECS433.pdf

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

QueenW1998 commented 2 years ago

师兄,您好! 我看了您这个文档的相关介绍,有几个问题请您帮我解答一下:

1、您在文档中介绍说这个程序只适合单人情景下行为动作识别,请问能提取多人状态下的骨骼信息吗,还是说只是检测行为的时候,因为您所使用的的数据集都是单人状态,所以只适合单人行为识别。如果我换成多人数据集训练,可以实现多人场景行为识别吗?

2、src\s4_train.py这个脚本训练的是动作分类模型吗?但是我在程序里面看到的关于加载的cmu模型,openpose网络需要训练吗?

3、项目中我没有找到DNN分类网络,麻烦师兄帮忙指点。

4、数据集文件夹的命名方式:jump_03-02-12-34-01-795,jump代表动作,后面的数字有什么特殊含义?

5、项目tf-pose-estimation为什么需要附在整个代码里面,与您的这个项目Realtime-Action-Recognition-master有什么直接关系吗?您这个项目里面所用到的openpose程序网络在哪个脚本里写的呢?

6、请问tf-pose-estimation\models\graph里面关于cmu和mobilenet模型是预训练模型吗?它和tf-pose-estimation\models\pretrained有什么区别呢?它们的作用分别是什么呢?

7、openpose网络里面用到的cmu模型主要的功能是什么呢?它只能实现单人单目的检测吗?

8、我可以采用什么评价指标来判断openpose检测的关节点是否准确呢,如果不准确的话,有什么改进的方法吗?

9、整个项目有没有用到跟踪算法呢?

10、如果我想换成自己的数据集,对动作识别算法有什么可以改进的建议吗?

恳请师兄帮忙解答,不胜感激!

------------------ 原始邮件 ------------------ 发件人: "felixchenfy/Realtime-Action-Recognition" @.>; 发送时间: 2022年5月15日(星期天) 凌晨2:09 @.>; @.**@.>; 主题: Re: [felixchenfy/Realtime-Action-Recognition] 您好!可以请教一下是否有相关文章或者资料对这个项目的介绍分析吗 (Issue #82)

流程/算法我写在README里了啊: https://github.com/felixchenfy/Realtime-Action-Recognition#1-algorithm https://github.com/felixchenfy/Data-Storage/blob/master/EECS-433-Pattern-Recognition/FeiyuChen_Report_EECS433.pdf

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>