CLASS Datasets and Software
This page lists a few of the publicly available datasets and software associated with CLASS. If you
can't find something here, it may be available on one of the partner's web sites.
-
Unsupervised Object Discovery - datasets, visual features, learning methods and evaluation software
used in our
joint evaluation of several methods for unsupervised learning of visual classes.
- Multiple Kernels for Image Classification
- data set and software relating to our work on efficient image classification using
Multiple Kernel Learning over rich local feature sets.
- Learning Visual Attributes
- data sets and software relating to our NIPS 2007 paper
Learning Visual Attributes.
- Animals with Attributes
- annotations, visual features and software associated with our CVPR 2009 paper
Learning
To Detect Unseen Object Classes by Between-Class Attribute Transfer.
- Methods
for structured output based object detection from Christoph
Lampert: Efficient Subwindow Retrieval
(ICCV'09), Structured Regression for Localization (ECCV'08), Efficient
Subwindow Search (CVPR'08), ...
- Tuwo
Computer Vision Library by Sebastian
Nowozin: feature sets, learning methods, data set utilities,
...
- Learning methods for structured feature spaces for object
recognition: Sequence
Boosting Toolbox and Graph Boosting
Toolbox for Matlab.
- Multiple
Instance Learning - Spider Matlab code
for Multiple Instance Learning by Peter Gehler.
- IKL
- Infinite and Multiple Kernel Learning algorithm for Matlab by
Peter Gehler and Sebastian Nowozin.
- Spatio-Temporal
Interest Point Detector - information and software relating to our
ECCV 2008 paper An
Efficient Dense and Scale-Invariant Spatio-Temporal Interest Point
Detector.
- Exemplar-based
Action Recognition - information and software relating to our BMVC
2009 paper
Exemplar-based Action Recognition in Video.
- HOLLYWOOD2
Human Actions and Scenes Dataset - data associated with our
CVPR 2009 paper Actions in
Context.
- Tools
for computing spatio-temporal gradient descriptors in videos -
c.f. our BMVC 2008 paper A
Spatio-Temporal Descriptor Based on 3D-Gradients.
- The following pages contain description, examples, publications,
datasets and software associated with our work on finding
characters in TV shows, naming them and estimating their
upper-body poses. Most of the examples are from the TV series
Buffy the Vampire Slayer.
- Automatic
Naming of Characters - datasets and software associated with the
initial version of our pipeline for naming characters in TV shows.
- Upper-body detector
- data set and software relating to upper-body-based character detection.
- Buffy Stickmen
- annotated data set and evaluation routines relating to 2D human pose estimation.
- Buffy pose classes
- annotated data set and publications relating to qualitative pose classification.
- A Names and Faces dataset based on INRIA's work on automatically naming faces in news images should be
available in fall 2009.
Contact: Bill Triggs (coordinator), Bill.Triggs@imag.fr,
phone +33 4 7651 4553
Laboratoire Jean Kuntzmann, 51 rue des Mathematiques,
38402 Saint Martin d'Heres, Grenoble, France
