如何使用opencv或android中的任何其他库实现对象检测 [英] How to implement object detection using opencv or any other library in android

问题描述

我想检测一个可以修复的对象并使用Android Camera拍摄该对象的图片。我搜索谷歌搜索它，发现openCV似乎是最佳结果，但问题是没有确切的帮助，如何去做。我已经测试了openCV包中可用的样本。像人脸检测这样的样本也不起作用。如果有人尝试过任何类似的东西，请帮助。我也找到了BoofCV，但是接下来是一个小测验。 OpenCV看起来是使用c / C ++开发的，我是新手。在这方面的任何输入都会有所帮助。

I want to detect an object which would be fixed and take a picture of that object using Android Camera. I searched google for it and found openCV which seems to be optimal result but the problem is there is no exact help for how to go about it. I have already tested the samples that are available in the openCV package. Also the samples like face detection are not working. If anyone has tried any such similar thing please help. I also found BoofCV but hw to go about it is a quiz. OpenCV looks to be developed using c/C++ for which I am new. Any input in this regard will be helpful.

推荐答案

祝福 - 图像分析不是一个有明确解决方案的东西，也不是可以快速修补的东西。



OpenCV有一个带脸部识别自学习的示例项目，但它需要教授，我不能说是好还是坏 - 我从事图像分析但是从不同的角度来看。



您应该找到一些关于图像处理的好书，如数字图像处理：PIKS Inside，第三版.William K. Pratt（ISBN： 0-471-37407-5）或图像处理手册，John C. Russ（ISBN：0-8493-7254-2）用这个字段来掌握主要问题。



另外，深入研究OpenCV文档，初步了解我花了将近一周的问题。



人脸识别特别是在人工智能方面有很深的根源，因为算法麦克风，确定性方法往往是不可靠的，因为处理的图像中的微小变化可能会完全误导算法。





编辑：抱歉，我在谈论人脸检测，你要求检测对象。



在这种情况下，它甚至比面部检测还要糟糕，因为你必须知道您正在寻找。开始实现对象识别的最佳方法是使用Sobel或Laplace算子将图像转换为灰度级并使用woprk来获取图像的边缘。从那时起，根据背景，类似物体，物体交汇点等区分对象是一个逻辑问题。



希望有所帮助，

Denis

Blessings - image analysis is not something for which there is a clear solution nor something that can be patched up quickly.

OpenCV has a sample project with autolearning for face recognition though, it needs to be taught and I can't say if it's good or bad - I work in image analysis but from a different perspective.

You should find some good books on Image Processing like "Digital Image Processing: PIKS Inside, Third Edition. William K. Pratt" (ISBN: 0-471-37407-5) or "The image processing handbook, John C. Russ" (ISBN: 0‑8493‑7254‑2) to get a hold of the main problematics with this field.

Also, delve deep into OpenCV documentation, to have a preliminary knowledge of the problem it took me almost a week.

Face recognition in particular has deep roots in Artificial Intelligence because algorithmic, deterministic approaches tend to be unreliable due to the sheer number of small variations in the image processed that can completely mislead the algorithms.


sorry, I was talking about face detection and you were asking for object detection.

In this case it's even worst than face detection because you have to know what you're looking for. The best way to start implementing object recognition is to turn the image in graylevel and woprk your way out with Sobel or Laplace operators to get the edges of the image. From then on it is a matter of logic to discriminate objects with respect to background, similar objects, junctions of objects and so on.

Hope to be of help,
Denis

这篇关于如何使用opencv或android中的任何其他库实现对象检测的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！