datasets

Pascal Objects Eye Tracking (POET) dataset (September 2014)

Dataset of eye tracking data on Pascal VOC 2012 trainval images.

It appeared in the ECCV 2014 paper.

ImageNet bounding boxes and segmentations (June 2014)

Project page with annotations for ~500K images of ImageNet.

These are the results of our CVPR 2012, ECCV 2012 and CVPR 2014 papers.

Synchronic Activities Stickmen V 1.0 (August 2012)

Dataset of annotated 2D human body part configurations (poses) in synchronic activity scenario.

It appeared in the PAMI 2012 paper.

Youtube-Objects dataset (June 2012)

Dataset of videos collected from YouTube by querying for the names of 10 object classes, with partial bounding-box annotations. Point tracks and motion segments as well as spatio-temporal bounding boxes are provided as well.

It appeared in the CVPR 2012 paper.

FAN-Large (September 2011)

Dataset of 125000 image-caption pairs.

It appeared in the BMVC 2011 paper.

We Are Family: Joint Pose Estimation of Multiple PersonsWe Are Family Stickmen V 1.02 (July 2012)

Dataset of group photos for human pose estimation with annotated 2D human body part configurations.

It appeared in our ECCV 2010 paper.

Idiap/ETHZ Faces and Poses V 1.0 (April 2010)

Dataset of image-caption pairs with famous persons performing various actions.

It appeared in the NIPS 2009 paper.

ETHZ PASCAL Stickmen V 1.11 (July 2012)

Dataset of annotated 2D human body part configurations (poses)

It appeared in the BMVC 2009, IJCV 2012 and PAMI 2012 papers.

Buffy Pose Classes V 1.0 (September 2009)

Dataset of video shots annotated by whether they contain any of three poses classes.

It appeared in the CVPR 2009 paper.

Buffy Stickmen V 3.01 (July 2012)

Dataset of annotated 2D human body part configurations (poses)

It appeared in the CVPR 2008, CVPR 2009, BMVC 2009, IJCV 2012 and PAMI 2012 papers

New in V 3.0 and 3.01:

Visual Attributes (December 2007)

Project page with our dataset for training and testing visual attribute recognition algorithms (e.g. red, striped).

It appeared in the NIPS 2007 paper.

INRIA Horses V 1.03 (November 2009)

Dataset for testing object class detection algorithms. It contains 170 images with horses and 170 images without.

All horses in all images are annotated with a bounding-box.

The package also includes edgemaps produced by the excellent Berkeley natural boundary detector.

This dataset appeared in the CVPR 2007, PAMI 2008, and IJCV 2010 papers.

Dowload the dataset

New in V 1.02 and 1.03:

ETHZ Shape Classes V 1.2 (September 2009)

Dataset for testing object class detection and shape matching algorithms.

It appeared in the ECCV 2006, CVPR 2007, PAMI 2008, and IJCV 2010 papers.

Download the entire ETHZ Shape Class Dataset

New in V 1.2: Alternative annotations by Xavier Giro:

You can download binary mask and alternative bounding-box annotations by Xavier Giro. These are useful for algorithms requiring a solid mask as input or for evaluation, rather than a contour. Important: this material is not part of the official protocol. If you want to compare to previous works, please use the annotations in the main package above.

Models

Apple Logo Bottle Giraffe Mug Swan

Test

Apple Logos
(40 images)
Bottles
(48 images)
Giraffes
(87 images)
Mugs
(48 images)
Swans
(32 images)

ETHZ Toys V 1.0

Dataset for testing algorithms for the recognition of specific objects.

It appeared in the ECCV 2004 and IJCV 2006 papers.

Download the entire ETHZ Toys Dataset

Models

Some objects in the dataset are imaged from multiple viewpoints. Browse All *.


(1 view)

(1 view)

(8 views)

(1 view)

(1 view)

(8 views)

(6 views)

(8 views)

(6 views)
 

*Note that the browsable images for this dataset are in jpeg format. Please download the full dataset to obtain the original tiff images.

Test

   

Browse Larger Thumbnails *.