使用java从图像系列/集合中进行人物检测
我收集了一个人不同角度的图像（可能是 10 张） 我需要查出这个人。
我们可以使用这些初始 10 张图像进行训练/特征提取 当另一个（第 11 个）图像作为输入给出时，我们需要检测该图像中的人，
我需要用 Java 编码 非常欢迎任何帮助/链接。 请帮忙。
谢谢。
I have collection of images(may be 10) of one person from different angles
I need to detect this person.
We can use these initial 10 images for training/feature extraction
and when another(11th) image is given as input, we need to detect person in that image
I need to code in Java
any help/links are most welcome.
Please help.
Thank you.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
为了正确地回答这个问题，我需要更多信息。您有兴趣专门检测那个人吗？该人是否有任何可以通过机器学习技术轻松利用的特征？您能获得更多训练数据吗？有什么办法可以看到图片吗？
首先：
您可以使用神经网络，但我怀疑您可能需要更多的训练数据来有效，无论如何肯定值得学习。
您还可以尝试由图像的不同部分组成的卷积蒙版（看看会很有趣哪些功能是最好的指标）。
您可以尝试使用霍夫变换（在维基百科上搜索）。如果您只是想检测图像中是否存在一个人（任何人），您可以寻找椭圆形。
To answer this question decently I'd need more information. Are you interested in detecting specifically that person? Are there any characteristics of that person that could be easily exploited with machine learning techniques? Are you able to get hold of any more training data? Is there any way I could see the pictures?
As a start:
You could use a neural net but I suspect you may need more training data for that to be effective, in any case certainly worth learning about.
You could also try convolution masks made up of different parts of the image (it would be interesting to see which features are the best indicator).
You could try using a Hough Transform (search for this on Wikipedia). If you are just interested in detecting whether a person (any person) is present in the image you could look for ovals.