My goal is to detect cars in images and recognize its model. For car-detection, from http://docs.opencv.org/trunk/opencv_cheatsheet.pdf, it says:
CascadeClassifier
Viola’s Cascade of Boosted classifiers us-ing Haar or LBP features.
Suits for de-tecting faces, facial features and some other objects
without diverse textures. See facedetect.cpp
HOGDescriptor
N. Dalal’s object detector using Histogram-of-Oriented-Gradients (HOG)
features. Suits for detect-ing people, cars and other objects with
well-defined silhouettes. See peopledetect.cpp
and from http://docs.opencv.org/opencv2refman.pdf , chapter 8.1 Cascade Classification ,page 421:
First, a classifier (namely a cascade of boosted classifiers working
with haar-like features) is trained with a few hundred sample views of
a particular object (i.e., a face or a car), called positive examples,
that are scaled to the same size (say, 20x20), and negative examples -
arbitrary images of the same size.
Both of these two methods mention about applying for car detection: in opencv_cheatsheet.pdf it says HOGDescriptor suits for cars and in opencv2refman.pdf it also mentions cars for Cascade Classification. So my question is how to choose between CascadeClassifier and HOGDescriptor, and which is better for car detection? Thank you.
Its hard to tell without testing, but actually I got very good results with a third option - latentSVM detector.
It depends on your goals in terms of speed, accuracy etc. You use cascade classifiers and can get the desired results. Although cascade classifiers are quite good but still they are computational expensive. In that case you can you gpu. Below is an few example for you.
https://github.com/abhi-kumar/CAR-DETECTION
It's a research question. Classifier trained by and CascadeClassifier and HOG can be used to detect vehicles. OpenCV provide a detailed help to build a customized detector in case of CascadeClassifier and HOG. Quality of the classifier can be described by accuracy and in my case I was more interested in real time implementation. So you have to look a number of different factors.
Related
I am conducting binary classification using logistic regression with and without applying PCA. The application of PCA before logistic regression gives a higher accuracy and lower FNs in comparison to logistic regression alone. I would like to find out why this is happening, specifically why PCA produces less FNs. I have read that cost sensitivity analysis could help explain this, but I am not sure if this is correct. Any suggestions?
There is no need of fancy analysis to explain this behavior.
PCA is used just for "clean" the data by limiting its variance. Let me explain this concept with an example, and then I will turn back to your question.
In general, in any ML problem, the available samples are never sufficient in number to cover all the possible variety of the sample space. You can never have a dataset with all the possible human faces, with all the possible expressions, etc.
So, instead of using all the available features you engineer the features (the pixels, in this example) in a way that you get more meaningful higher level features. You can reduce the resolution of the pictures, as easy example; you will loose the informations on the pictures background, but your model will focus better on the most important part of the picture, i.e. the faces.
When you deal with tabular data, a technique similar to the resolution lowering is cutting off parts of the original features, and that's what PCA do: it keeps the most important components of the features, the "Principal Components", dropping the less important ones.
So, the model trained with PCA gives better results because, by cutting off part of the features, your model focus better on the most important part of your samples, and so it gains robustness against overfitting.
cheers
I am using the Caffe framework for CNN training. My aim is to perform simple object recognition for a few basic object categories. Since pretrained networks are not an alternative for my proposed usage I prepared an own training- and testset with about 1000 images for each of 2 classes (say chairs and cars).
The results are quite good. If I present an yet unseen image of a chair it is likely classified as such, same for a car image. My problem is that the results on miscellaneous images that do not show any of these classes often shows a very high confidence (=1) for one random class (which is not surprising regarding the onesided training data but a problem for my application). I thought about different solutions:
1) Adding a third class with also about 1000 negative examples that shows any objects except a chair and a car.
2) Adding more object categories in general, just to let the network classify other objects as such and not any more as a chair or car (of course this would require much effort). Maybe also the broader prediction results would show a more uniform distribution at negative images, allowing to evaluate the target objects presence based on a threshold?
Because it was not much time-consuming to grab random images as negative examples from the internet, I already tested my first solution with about 1200 negative examples. It helped, but the problem remains, perhaps because it were just too few? My concern is that if I increment the number of negative examples, the imbalance of the number of examples for each class leads to less accurate detection of the original classes.
After some research I found one person with a similar problem, but there was no solution:
Convolutional Neural Networks with Caffe and NEGATIVE IMAGES
My question is: Has anyone had the same problem and knows how to deal with it? What way would you recommend, adding more negative examples or more object categories or do you have any other recommendation?
The problem is not unique to Caffe or ConvNets. Any Machine Learning technique runs this risk. In the end, all classifiers take a vector in some input space (usually very high-dimensional), which means they partition that input space. You've given examples of two partitions, which helps to estimate the boundary between the two, but only that boundary. Both partitions have very, very large boundaries, precisely because the input space is so high-dimensional.
ConvNets do try to tackle the high-dimensionality of image data by having fairly small convolution kernels. Realistic negative data helps in training those, and the label wouldn't really matter. You could even use the input image as goal (i.e. train it as an autoencoder) when training the convolution kernels.
One general reason why you don't want to lump all counterexamples is because they may be too varied. If you have a class A with some feature value from the range [-1,+1] on some scale, with counterexamples B [-2,-1] and C [+1,+2], lumping B and C together creates a range [-2,+2] for counterexamples which overlaps the real real range. Given enough data and powerful enough classifiers, this is not fatal, but for instance an SVM can fail badly on this.
I am trying to detect the vehicles for that i used the SURF/SIFT and BOW with SVM approach but vehicles are of different type and i just studies that SURF/SIFT is for one specific object detection like usb, mobile phone etc. Is that also mean that it also affect (in detection) different types of car's like toyota and bmw etc ? or vehicles like truck and car ?
If we provide a large number of dataset of 10/15 different vehicles to SURF/SIFT then isn't it give the decent result with detection of different type of vehicles using real time approach?
SURF/SIFT are spatial local features. If the dataset is large enough the result should be good. Even for different vehicles only specific structures of that vehicle are available in a scene image.
However false positives might creep in if similar non-vehicle structure is present. (E.g distorted image of a small rectangular house with fence). So, some global feature like road detection might increase accuracy.
So, i think sirf/surf features of vehicles with single class SVM should help if the false positives are not present in the image of your application.
It seems that features and possibly a processing method are chosen incorrectly. Features, for example, should be representative for all classes of cars and thus cannot be particular interest points with descriptors representing just some peculiarities of gradient around the point and matching looking for exact spacial configuration of points relative to other points.
The processing with SVM means you classify cars against all other objects. I am not sure how you are going to get “all other objects” supporting vectors though. Much more sensible features are the same the humans seems to use to detect cars, try HOG paper for example that uses mixture of deformable parts. This closer to the sate of the art and got multiple awards so there is no need to invent a bicycle.
I am tasked with detecting anomalies (known or unknown) using machine-learning algorithms from data in various formats - e.g. emails, IMs etc.
What are your favorite and most effective anomaly detection algorithms?
What are their limitations and sweet-spots?
How would you recommend those limitations be addressed?
All suggestions very much appreciated.
Statistical filters like Bayesian filters or some bastardised version employed by some spam filters are easy to implement. Plus there are lots of online documentation about it.
The big downside is that it cannot really detect unknown things. You train it with a large sample of known data so that it can categorize new incoming data. But you can turn the traditional spam filter upside down: train it to recognize legitimate data instead of illegitimate data so that anything it doesn't recognize is an anomaly.
There are various types of anomaly detection algorithms, depending on the type of data and the problem you are trying to solve:
Anomalies in time series signals:
Time series signals is anything you can draw as a line graph over time (e.g., CPU utilization, temperature, rate per minute of number of emails, rate of visitors on a webpage, etc). Example algorithms are Holt-Winters, ARIMA models, Markov Models, and more. I gave a talk on this subject a few months ago - it might give you more ideas about algorithms and their limitations.
The video is at: https://www.youtube.com/watch?v=SrOM2z6h_RQ
Anomalies in Tabular data: These are cases where you have feature vector that describe something (e.g, transforming an email to a feature vector that describes it: number of recipients, number of words, number of capitalized words, counts of keywords, etc....). Given a large set of such feature vectors, you want to detect some that are anomalies compared to the rest (sometimes called "outlier detection"). Almost any clustering algorithm is suitable in these cases, but which one would be most suitable depends on the type of features and their behavior -- real valued features, ordinal, nominal or anything other. The type of features determine if certain distance functions are suitable (the basic requirement for most clustering algorithms), and some algorithms are better with certain types of features than others.
The simplest algo to try is k-means clustering, where an anomaly sample would be either very small clusters or vectors that are far from all cluster centers. One sided SVM can also detect outliers, and has the flexibility of choosing different kernels (and effectively different distance functions). Another popular algo is DBSCAN.
When anomalies are known, the problem becomes a supervised learning problem, so you can use classification algorithms and train them on the known anomalies examples. However, as mentioned - it would only detect those known anomalies and if the number of training samples for anomalies is very small, the trained classifiers may not be accurate. Also, because the number of anomalies is typically very small compared to "no-anomalies", when training the classifiers you might want to use techniques like boosting/bagging, with over sampling of the anomalies class(es), but optimize on very small False Positive rate. There are various techniques to do it in the literature --- one idea that I found to work many times very well is what Viola-Jones used for face detection - a cascade of classifiers. see: http://www.vision.caltech.edu/html-files/EE148-2005-Spring/pprs/viola04ijcv.pdf
(DISCLAIMER: I am the chief data scientist for Anodot, a commercial company doing real time anomaly detection for time series data).
I have been playing around with different data clustering algorithms working on finding clusters between random data points represented an nodes, I keep reading that data clustering is used for image recognition. I am failing to make the connection, how does clustering data help in recognizing an image or in facial recognition. can someone explain this?
It's no surprise that clustering is used for pattern recognition at large, and image recognition in particular: clustering is a reducing process, and images in this megapixel era need boiling down... It is also a process which produces categories and that is of course useful.
However there are many approaches to the use of clustering as a technique for image recognition. One of the reasons for this diversity is that clustering can be applied at different level, for different purposes: from basic pixel level to feature level (feature be a line, a geometric figure...), for classification or for other purposes.
At a very high level, clustering is a statistical tool, it helps discovering the relative importance of various dimensions in defining the belonging of particular item to a particular category.
One [of many] usage[s] of such a tool, is with supervised learning, whereby a set of human-selected items (say images) are fed into the cluster-based logic, along with a label associated with a particular item ("this is an apple", "this is another apple", "this is a lemon"...), the clustering logic then determines how much each dimension of the input matters for helping each group of items (apples, lemons...) fit in a distinct cluster (for example the color may matter relatively little, but the shape, or the presence of dots, or whatever may matter a lot). After this training phase, new images can be fed to the logic and by seeing how close to a particular cluster this image falls, it is "recognized" (as a banana!).
When it comes to image processing one needs to remember that whatever is "fed" to the clustering logic is not necessarily (in fact, rarely) the raw pixels, but various "objects"
characterizing various "elements" of the original data (essentially a collection of relatively high dimension vectors, not unlike some that one may have encountered in other other data clustering examples), and produced by previous stages of the process. For example a important element of facial recognition is probably the exact distance between the center of the eyes. In previous stages, the image is processed in a way that figures out where the eyes are (possibly relying on another clustering-based logic). Then the distance between the eyes, along with many other elements are fed to the final clustering logic.
The preceding description is only one example of the use of clustering for image recognition. Indeed, various forms of neural networks have been used, very successfully, in this domain, and it can be argued that in a sense these neural networks are clustering information. One of the reasons for the success of neural nets may lie in their ability to be more respectful of the locality dimension as found in the original input, and also their ability to work in a hierarchical fashion.
A good conclusion to this write up would be a short list of online resources, but I'm pressed for time at the moment... "to be continued" ;-)
Next day edit: (failed attempt to provide an introductory online bibliography on the subject)
My search for literature on the topic of clustering as applied to artificial vision and image processing revealed two distinct... clusters ;-)
Books such as Algorithms for image processing and computer vision J Parkey pub Wiley, or Machine Vision : Theory, Algorithms, Practicalities M Seul et. Al Cambridge UP. Such books generally cover the all important techniques associated with noise reduction, Edge detection, Color or intensity conversion, and many other elements of the image processing chain, most of which do not involve clustering or even statistical methods, and they reserve only a chapter or two, or even minor mentions, to clustering, as applied to pattern recognition or to other tasks.
Scholarly papers and conference handbooks, which specifically cover clustering techniques applied to artificial vision and such, but in the narrowest and deepest fashion (ex: Variations on the Fukunaga and Narendra algorithm, for applications in character recognition, or Fast methods for selections of Nearest Neighbor candidates in whatever context.)
In short I feel ill equipped to make any specific book or article suggestion.
You may find it informative to browse titles in say Google books, keying in by "Artificial vision" or "Image Recognition" or some or the titles mentioned above. With the preview feature and also the tag cloud (btw another application of clustering) found in the "about this book" link, one can get a good idea of the various books contents and maybe decide to purchase some of them. Unfortunately the reduced readership and the potentially lucrative applications in the field make these books relatively expensive. At the other end of the spectrum, you may download, sometimes for free, research papers discussing advanced topics in the field. These will also show up on regular (web) Google, or at specialized repositories such as CiteSeer.
Good luck with your exploration in that field!