• Corpus ID: 18859874

Speeded-up and Compact Visual Codebook for Object Recognition

  title={Speeded-up and Compact Visual Codebook for Object Recognition},
  author={Amirthalingam Ramanan and Sinnathamby Mahesan and U. A. J. Pinidiyaarachchi},
The well known framework in the object recognition literature uses local information extracted at several patches in images which are then clustered by a suitable clustering technique. A visual codebook maps the patch-based descriptors into a fixed-length vector in histogram space to which standard classifiers can be directly applied. Thus, the construction of a codebook is an important step which is usually done by cluster analysis. However, it is still difficult to construct a compact… 

Figures and Tables from this paper

Compact codebook design for visual scene recognition by Sequential Input Space Carving
The Sequential Input Space Carving (SISC) approach is presented, which achieves compact codebooks in a fraction of the computation time needed by the k-means clustering method usually employed in this setting.
An overview and categorization of the approaches used in the literature for the purpose of Generic Object Recognition and various technical advancements achieved in the field are given.


A one-pass resource-allocating codebook for patch-based visual object recognition
  • A. Ramanan, M. Niranjan
  • Computer Science
    2010 IEEE International Workshop on Machine Learning for Signal Processing
  • 2010
This work extends recent work on constructing a one-pass discriminant codebook design procedure inspired by the resource allocating network model from the artificial neural networks literature and retains data spread out more widely in the input space, thereby including rare low-level features in the codebook.
Creating efficient codebooks for visual recognition
  • F. Jurie, B. Triggs
  • Computer Science
    Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1
  • 2005
It is shown that dense representations outperform equivalent keypoint based ones on these tasks and that SVM or mutual information based feature selection starting from a dense codebook further improves the performance.
Learning Optimal Compact Codebook for Efficient Object Categorization
A novel approach of learning optimal compact codebooks by selecting a subset of discriminative codes from a large codebook, where the Gaussian models of object categories based on a single code are learned from the distribution of local features within each image.
Compact visual codebook for action recognition
This paper proposes a novel method to address the problem of ambiguity and weak efficiency in the visual codebook by constructing a compact but effective visual codebooks using sparse reconstruction, and applies it to action recognition tasks.
Category sensitive codebook construction for object category recognition
A novel optimization method for discriminative codebook construction that considers the category information of local image features as an additional term in traditional visual-similarity-only based codebookConstruction methods is proposed.
Large-scale robust visual codebook construction
An improved approximate K-means is proposed, by leveraging the assignment information in the history, namely the previous iterations, to improve the assignment precision by further randomizing the employed approximate nearest neighbor search in each iteration.
Object categorization by learned universal visual dictionary
An optimally compact visual dictionary is learned by pair-wise merging of visual words from an initially large dictionary, and a novel statistical measure of discrimination is proposed which is optimized by each merge operation.
SVM-KNN: Discriminative Nearest Neighbor Classification for Visual Category Recognition
This work considers visual category recognition in the framework of measuring similarities, or equivalently perceptual distances, to prototype examples of categories and proposes a hybrid of these two methods which deals naturally with the multiclass setting, has reasonable computational complexity both in training and at run time, and yields excellent results in practice.
Universal and Adapted Vocabularies for Generic Visual Categorization
  • F. Perronnin
  • Computer Science
    IEEE Transactions on Pattern Analysis and Machine Intelligence
  • 2008
This work proposes a novel practical approach to GVC based on a universal vocabulary, which describes the content of all the considered classes of images, and class vocabularies obtained through the adaptation of the universal vocabulary using class-specific data.
Speeding up multi-class texture classification by one-pass vocabulary design and decision tree
This paper presents a fast approach for texture classification that integrates existing ideas to relieve the excessive time involved both in constructing a visual vocabulary and classifying unknown images using a support vector machine based decision tree.