Incremental learning of object detectors using a visual shape alphabet

By Andreas Opelt, Axel Pinz, Andrew Zisserman. 2006

206 citations recorded by the source corpus.

We address the problem of multiclass object detection. Our aims are to enable models for new categories to benefit from the detectors built previously for other categories, and for the complexity of the multiclass system to grow sublinearly with the number of categories. To this end we introduce a visual alphabet representation which can be learnt incrementally, and explicitly shares boundary fragments (contours) and spatial configurations (relation to centroid) across object categories. We develop a learning algorithm with the following novel contributions: (i) AdaBoost is adapted to learn jointly, based on shape features; (ii) a new learning schedule enables incremental additions of new categories; and (iii) the algorithm learns to detect objects (instead of categorizing images). Furthermore, we show that category similarities can be predicted from the alphabet. We obtain excellent experimental results on a variety of complex categories over several visual aspects. We show that the sharing of shape features not only reduces the number of features required per category, but also often improves recognition performance, as compared to individual detectors which are trained on a per-class basis.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Low-Shot Learning from Imaginary Data (2018, background): “Some work tries to share features between seen and novel classes [1, 36] or incrementally learn them as new classes are encountered [21].”
Incremental Few-Shot Object Detection (2020, neutral): “(1)Some works have studied incremental learning of object detectors [35].”
Learning a category independent object detection cascade (2011, methodology): “made use of a shared shape alphabet to reduce the complexity of object detection [18].”
50 Years of object recognition: Directions forward (2013, methodology): “Their work is further expanded upon in [196].”
An Efficient Shared Multi-Class Detection Cascade (2008, background): “Obviously, such solutions can be extended to multi-class detection, i.e. determining where in the images is any of the target classes of objects, by repeating a search for each of those classes separately.”
Low-Shot Visual Recognition by Shrinking and Hallucinating Features (2016, unknown): “Among discriminative approaches, early work attempted to use a single image of the novel class to adapt classifiers from similar base classes [4, 31] using simple hand-crafted features.”
Representing shape with a spatial pyramid kernel (2007, background): “Third, we show that shape and appearance kernels may be combined (again by learning parameters on a validation set).”
Pedestrian Detection via Classification on Riemannian Manifolds (2008, unknown): “Ç”
Learning to share visual appearance for multiclass object detection (2011, background): “Other models learn the relations between categories [30, 21, 22, 1, 25, 13, 29].”
Learning the Compositional Nature of Visual Objects (2007, background): “Opelt et al . of curves and their relative position to the object center.”
Shape Sharing for Object Segmentation (2012, unknown): “In object detection, jointly training multi-class detectors allows the reuse of common discriminative features [13, 14].”
A Review of Computer Vision Techniques for the Analysis of Urban Traffic (2011, background): “An extension to multiple classes and incremental learning is introduced in [105] and [107].”
Learning to Learn: Model Regression Networks for Easy Small Sample Learning (2016, background): “In particular, inter-class transfer [40] and crossgeneralization [7] are achieved by discovering shared feature representations: 1) captured by linear or nonlinear feature transformations [14, 63, 85, 31, 1, 94, 48], 2) obtained by feature selection [59, 60, 27] or regularization [37], 3) described ”
Multi-Class Segmentation with Relative Location Prior (2008, background): “, [17, 30]), multi-class image segmentation methods are aimed at concurrent multiclass object recognition and attempt to classify all pixels in an image (e.”
An Exemplar Model for Learning Object Classes (2007, methodology): “We are bringing together three previous ideas: (i) the generalized Hough Implicit Shape model of Leibe et al. [14] using sparse appearance patches, (ii) the edge/boundary representation of [17, 18 , 20], and (iii) correspondence matching using a hierarchical spatial histogram [10, 13].”
Low-shot visual object recognition (2016, unknown): “Among discriminative approaches, early work attempted to use a single image of the novel class to adapt classifiers from similar base classes [3, 28], but only used simple hand-crafted features.”
Human Detection via Classification on Riemannian Manifolds (2007, methodology): “Other approaches include using silhouette information either in matching [8] or in classification framework [15].”