jasonleaster / facedetection Goto Github PK

Constructing AdaBoost from existed model data
Construction finished

Traceback (most recent call last):
File "C:/Users/rpa-s/Desktop/FaceDetection-master/FaceDetection/EFace.py", line 27, in
rectangles = det.scanImgOverScale(imgSingleChannel)
File "C:\Users\rpa-s\Desktop\FaceDetection-master\FaceDetection\detector.py", line 114, in scanImgOverScale
rectangles += self.scanImgAtScale(model, image, scale)
File "C:\Users\rpa-s\Desktop\FaceDetection-master\FaceDetection\detector.py", line 54, in scanImgAtScale
subImages[idx] = Image(Mat = image[y:y+SEARCH_WIN_HEIGHT, x:x+SEARCH_WIN_WIDTH])
File "C:\Users\rpa-s\Desktop\FaceDetection-master\FaceDetection\image.py", line 33, in init
assert Mat != None
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

我更改了EFace中的TEST_IMG的路径到指定的文件上，运行之后出现了请问这个报错应该怎么处理...

你好，感谢你的项目。我在使用提供的model测试图片时，得到的结果

请问是哪里出了问题吗？

If DEBUG_MODEL is True, the following code in cascade.py will crash:
self._mat = numpy.zeros((self.haar.featuresNum, tot_samples))

No such file or directory: './features/features_train.cache.npy'

how to fix it?

你好！
请问一下，Adaboost.py文件，Adaboost类中，def train(self)方法中，有这样一句。

    for m in xrange(self.weakerLimit):
        self.N += 1

        if DEBUG_MODEL == True:
            weaker_start_time = time.time()

        self.G[m] = self.Weaker(self._mat, self._label, self.W)
        
        errorRate = self.G[m].train()

请问一下 self.G[m] = self.Weaker(self._mat, self._label, self.W)，这样的话不是G[0],G[1],G[2],G[3]都是同一个弱分类器了吗？因为后面的参数都是一样的。谢谢！！！

Thanks for your great work
Have you finished Cascade train?

作者
不好意思 現在我運行 trainingAdaboost.py 的時候已經可以訓練也保存完成了
但是在執行EFace.py的時候會產生 ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() 這樣的error
請問作者也有發生過同樣的問題？

你好，在实现弱分类器的时候，你用(miuPos + miuNeg)/2作为弱分类器的阈值，这个阈值是一个合理的选择吗？我修改了你的这部分代码，在特征里找了一些离散值来寻找最优的阈值，但是训练的时候选择到的特征和论文描述不符合

https://github.com/bruceyang2012/Face-Alignment-with-simple-cnn

请问vecRectSum中idxVector具体含义是什么？idx1,idx2,idx3,idx4代表的是什么呢？idx1,idx2,idx3,idx4计算式是何含义?
希望大神能抽空解除下我的疑惑，万分焦急！万分感谢！

With the limitation of my own PC, I have used out of all training set which I have to build this face detection system -- EFace. If you have more powerful machine and be interesting in how to build a face detection system, just email me : )

Hello @jasonleaster ,
It looks like that you didn't finish the part of Cascade of stronger classifier at this project you shared. So have you complete it later. Or can you share it .

Thanks a lot!!!

Like this . I know i have the train datas, so it should build the featuretrain_cache.npy. But none. ido't understand why