Object Detection Using the Statistics of Parts

Abstract

Abstract. In this paper we describe a trainable object detector and its instantiations for detecting faces and cars at any size, location, and pose. To cope with variation in object orientation, the detector uses multiple classifiers, each spanning a different range of orientation. Each of these classifiers determines whether the object is present at a specified size within a fixed-size image window. To find the object at any location and size, these classifiers scan the image exhaustively.Each classifier is based on the statistics of localized parts. Each part is a transform from a subset of wavelet coefficients to a discrete set of values. Such parts are designed to capture various combinations of locality in space, frequency, and orientation. In building each classifier, we gathered the class-conditional statistics of these part values from representative samples of object and non-object images. We trained each classifier to minimize classification error on the training set by using Adaboost with Confidence-Weighted Predictions . In detection, each classifier computes the part values within the image window and looks up their associated classconditional probabilities. The classifier then makes a decision by applying a likelihood ratio test. For efficiency, the classifier evaluates this likelihood ratio in stages. At each stage, the classifier compares the partial likelihood ratio to a threshold and makes a decision about whether to cease evaluation-labeling the input as non-object-or to continue further evaluation. The detector orders these stages of evaluation from a low-resolution to a high-resolution search of the image. Our trainable object detector achieves reliable and efficient detection of human faces and passenger cars with out-of-plane rotation.

A survey on face detection in the wild: Past, present and future

“…Here we survey some part-based methods, such as [49] and [50], that do not fall in the above described DPM architecture and have attracted a lot of attention.…”

“…This family of algorithms mainly revolves around extensions and variations of the general object detection methodology [35,47]. Other notable parts-based methods include [49,50,51].…”

Learning Statistical Structure for Object Detection

“…The A human eye detector trained by this method has been tested extensively. The eyes were successfully located within a radius of 15 pixels with an accuracy of 98.2% on over 29,000 images of faces in an experiment independently conducted at the National Institute of Standards and Technology (NIST) by NIST employees 1 and reported back to the author. This dataset is sequestered and is not available to the public.…”

“…We measure performance by the area under the receiver operating characteristic (ROC) [13]. This measure of classification error accounts for the classifier's full operating range over values for the threshold, λ, in equation (1).…”

“…Sparse structuring of statistical dependency explains the empirical success of "parts-based" methods for face detection and face recognition [1][2] [3][4] [5]. Such parts-based methods concentrate modeling power on localized regions, in which dependencies tend to be strong and use weaker models over larger areas in which dependencies tend to be less significant.…”