Bill Triggs

Histograms of oriented gradients for human detection

We study the question of feature sets for robust visual object recognition; adopting linear SVM based human detection as a test case. After reviewing existing edge and gradient based descriptors, we show experimentally that grids of histograms of oriented gradient (HOG) descriptors significantly outperform existing feature sets for human detection. We study the influence of each stage of the computation on performance, concluding that fine-scale gradients, fine orientation binning, relatively coarse spatial binning, and high-quality local contrast normalization in overlapping descriptor blocks are all important for good results. The new approach gives near-perfect separation on the original MIT pedestrian database, so we introduce a more challenging dataset containing over 1800 annotated human images with a large range of pose variations and backgrounds.

Bundle Adjustment - A Modern Synthesis

This paper is a survey of the theory and methods of photogrammetric bundle adjustment, aimed at potential implementors in the computer vision community. Bundle adjustment is the problem of refining a visual reconstruction to produce jointly optimal structure and viewing parameter estimates. Topics covered include: the choice of cost function and robustness; numerical optimization including sparse Newton methods, linearly convergent approximations, updating and recursive methods; gauge (datum) invariance; and quality control. The theory is developed for general robust cost functions rather than restricting attention to traditional nonlinear least squares.

Human detection using oriented histograms of flow and appearance

Detecting humans in films and videos is a challenging problem owing to the motion of the subjects, the camera and the background and to variations in pose, appearance, clothing, illumination and background clutter. We develop a detector for standing and moving people in videos with possibly moving cameras and backgrounds, testing several different motion coding schemes and showing empirically that orientated histograms of differential optical flow give the best overall performance. These motion-based descriptors are combined with our Histogram of Oriented Gradient appearance descriptors. The resulting detector is tested on several databases including a challenging test set taken from feature films and containing wide ranges of pose, motion and background variations, including moving cameras and backgrounds. We validate our results on two challenging test sets containing more than 4400 human examples. The combined detector reduces the false alarm rate by a factor of 10 relative to the best appearance-based detector, for example giving false alarm rates of 1 per 20,000 windows tested at 8% miss rate on our Test Set 1.

Sampling strategies for bag-of-features image classification

Bag-of-features representations have recently become popular for content based image classification owing to their simplicity and good performance. They evolved from texton methods in texture analysis. The basic idea is to treat images as loose collections of independent patches, sampling a representative set of patches from the image, evaluating a visual descriptor vector for each patch independently, and using the resulting distribution of samples in descriptor space as a characterization of the image. The four main implementation choices are thus how to sample patches, how to describe them, how to characterize the resulting distributions and how to classify images based on the result. We concentrate on the first issue, showing experimentally that for a representative selection of commonly used test databases and for moderate to large numbers of samples, random sampling gives equal or better classifiers than the sophisticated multiscale interest operators that are in common use. Although interest operators work well for small numbers of samples, the single most important factor governing performance is the number of patches sampled from the test image and ultimately interest operators can not provide enough patches to compete. We also study the influence of other factors including codebook size and creation method, histogram normalization method and minimum scale for feature extraction.

Recovering 3D human pose from monocular images

We describe a learning-based method for recovering 3D human body pose from single images and monocular image sequences. Our approach requires neither an explicit body model nor prior labeling of body parts in the image. Instead, it recovers pose by direct nonlinear regression against shape descriptor vectors extracted automatically from image silhouettes. For robustness against local silhouette segmentation errors, silhouette shape is encoded by histogram-of-shape-contexts descriptors. We evaluate several different regression methods: ridge regression, relevance vector machine (RVM) regression, and support vector machine (SVM) regression over both linear and kernel bases. The RVMs provide much sparser regressors without compromising performance, and kernel bases give a small but worthwhile improvement in performance. The loss of depth and limb labeling information often makes the recovery of 3D pose from single silhouettes ambiguous. To handle this, the method is embedded in a novel regressive tracking framework, using dynamics from the previous state estimate together with a learned regression value to disambiguate the pose. We show that the resulting system tracks long sequences stably. For realism and good generalization over a wide range of viewpoints, we train the regressors on images resynthesized from real human motion capture data. The method is demonstrated for several representations of full body pose, both quantitatively on independent but similar test data and qualitatively on real image sequences. Mean angular errors of 4-6/spl deg/ are obtained for a variety of walking motions.

Autocalibration and the absolute quadric

The author describes a new method for camera autocalibration and scaled Euclidean structure and motion, from three or more views taken by a moving camera with fixed but unknown intrinsic parameters. The motion constancy of these is used to rectify an initial projective reconstruction. Euclidean scene structure is formulated in terms of the absolute quadric-the singular dual 3D quadric (4/spl times/4 rank 3 matrix) giving the Euclidean dot-product between plane normals. This is equivalent to the traditional absolute conic but simpler to use. It encodes both affine and Euclidean structure, and projects very simply to the dual absolute image conic which encodes camera calibration. Requiring the projection to be constant gives a bilinear constraint between the absolute quadric and image conic, from which both can be recovered nonlinearly from m/spl ges/3 images, or quasi-linearly from m/spl ges/4. Calibration and Euclidean structure follow easily. The nonlinear method is stabler, faster, more accurate and more general than the quasi-linear one. It is based on a general constrained optimization technique-sequential quadratic programming-that may well be useful in other vision problems.

Enhanced local texture feature sets for face recognition under difficult lighting conditions

Recognition in uncontrolled situations is one of the most important bottlenecks for practical face recognition systems. We address this by combining the strengths of robust illumination normalization, local texture based face representations and distance transform based matching metrics. Specifically, we make three main contributions: (i) we present a simple and efficient preprocessing chain that eliminates most of the effects of changing illumination while still preserving the essential appearance details that are needed for recognition; (ii) we introduce Local Ternary Patterns (LTP), a generalization of the Local Binary Pattern (LBP) local texture descriptor that is more discriminant and less sensitive to noise in uniform regions; and (iii) we show that replacing local histogramming with a local distance transform based similarity metric further improves the performance of LBP/LTP based face recognition. The resulting method gives state-of-the-art performance on three popular datasets chosen to test recognition under difficult illumination conditions: Face Recognition Grand Challenge version 1 experiment 4, Extended Yale-B, and CMU PIE.

3D human pose from silhouettes by relevance vector regression

We describe a learning based method for recovering 3D human body pose from single images and monocular image sequences. Our approach requires neither an explicit body model nor prior labelling of body pans in the image. Instead, it recovers pose by direct nonlinear regression against shape descriptor vectors extracted automatically from image silhouettes. For robustness against local silhouette segmentation errors, silhouette shape is encoded by histogram-of-shape-contexts descriptors. For the main regression, we evaluate both regularized least squares and relevance vector machine (RVM) regressors over both linear and kernel bases. The RVMs provide much sparser regressors without compromising performance, and kernel bases give a small but worthwhile improvement in performance. For realism and good generalization with respect to viewpoints, we train the regressors on images resynthesized from real human motion capture data, and test it both quantitatively on similar independent test data, and qualitatively on a real image sequence. Mean angular errors of 6-7 degrees are obtained - a factor of 3 better than the current state of the art for the much simpler upper body problem.

Autocalibration from Planar Scenes

This paper describes the theory and a practical algorithm for the autocalibration of a moving projective camera, from m ≥ 5 views of a planar scene. The unknown camera calibration, motion and scene geometry are recovered up to scale, from constraints encoding the motion-invariance of the cameras internal parameters. This extends the domain of autocalibration from the classical non-planar case to the practically common planar one, in which the solution can not be bootstrapped from an intermediate projective reconstruction. It also generalizes Hartleys method for the internal calibration of a rotating camera, to allow camera translation and to provide 3D as well as calibration information. The basic constraint is that orthogonal directions (points at infinity) in the plane must project to orthogonal directions in the calibrated images. Abstractly, the planes two circular points (representing its Euclidean structure) lie on the 3D absolute conic, so their projections must lie on the absolute image conic (representing the camera calibration). The resulting algorithm optimizes this constraint numerically over all circular points and all projective calibration parameters, using the inter-image homographies as a projective scene representation.

Face recognition based on image sets

We introduce a novel method for face recognition from image sets. In our setting each test and training example is a set of images of an individuals face, not just a single image, so recognition decisions need to be based on comparisons of image sets. Methods for this have two main aspects: the models used to represent the individual image sets; and the similarity metric used to compare the models. Here, we represent images as points in a linear or affine feature space and characterize each image set by a convex geometric region (the affine or convex hull) spanned by its feature points. Set dissimilarity is measured by geometric distances (distances of closest approach) between convex models. To reduce the influence of outliers we use robust methods to discard input points that are far from the fitted model. The kernel trick allows the approach to be extended to implicit feature mappings, thus handling complex and nonlinear manifolds of face images. Experiments on two public face datasets show that our proposed methods outperform a number of existing state-of-the-art ones.