a dataset for testing object class detection algorithms from ETH Zurich CV lab
Some object classes, by their nature, are better represented by contour features than by image patches or interest points. E.G. Mug
ETHZ Shape Classes
contains images of five diverse shape-based classes, collected from Flickr and Google Images. The images represent the clutter, intra-class shape variability, and scale changes.
- The dataset tries to include objects appearing at a wide range of scales. For example, object comprises only a rather small portion of the image.
- The objects are mostly unoccluded and are all taken from approximately the same viewpoint (the side).
The dataset has been collected and annotated by Vittorio Ferrari, and experiments on it first appeared in . They tackled the challenge of detecting objects in real images given a single hand-drawn example as ‘model’, the hand-drawings are included in release of Version: 1.2.
- Detect objects with hand-drawing: Using all 255 images as test set for every class. Hence, to search for one object, e.g., images of mug, making for a large negative test set. This is important as it allows to get a reliable value for the incidence of false-positives generated by the detection algorithm.
- Train model from real images: This dataset is also suited for the conventional setting in which models are learnt from real images (for example, by splitting the dataset in half training / half testing). The further results in this setting in [2,3,4]. Moreover [3,4] also report experiments in the setting of , i.e. using a single hand-drawn example as a model.
Five classes are covered(in total 255 images, 289 instances):
- apple logos, 40 images, 44 instances
- bottles, 48 images, 55
- giraffes, 87, 91
- mugs, 48, 66
- swans, 32, 33
Most images contain a single instance of an object class, while some contain multiple instances. No image contains instances of different classes.
Object bounding boxes are included in files
- Each line in the file encodes the bounding-box of an instance of <class> in <image>
- The coordinates of the bounding-box appear in the following format
top_left_x top_left_y bottom_right_x bottom_right_y
Groundtruth outlines(NOT used during training)
- Object outlines for applelogos, bottles, and giraffes are included in files (a separate file per object instance ):
- For mugs and swans, the outlines are in files(All instances are in the same file, and different instances have different greylevels):
The complete detection-rate vs FPPI performance plots for all their works [1,2,3,4], as they appeared in , are included in this release (as well as plots for the Chamfer Matching baseline).
plots in his directory correspond to figure 12 of . The models used for these plots have been trained from a subset of the ETHZ Shape Classes. The test images are a disjoint subset of thedataset.
plots in this directory correspond to figure 17 of . The models used for these plots the hand-drawings from the ETHZ Shape Classes. The test set are all real images in the ETHZ Shape Classes.
Refer to  for details of the meaning of each curve, and for the exact experimental setup.
In files *_edges.tif, edge maps produced by the excellent Berkeley ‘natural boundary detector’ were included. Using this advanced edge detector instead of the standard Canny, resulted in a significant improvement in object detection performance. We recommend using these edge maps.
 Vittorio Ferrari, Tinne Tuytelaars and Luc Van Gool, Object Detection by Contour Segment Networks, ECCV 2006, Graz, Austria
 Vittorio Ferrari, Loic Fevrier, Frederic Jurie, and Cordelia Schmid, Groups of Adjacent Segments for Object Detection,
PAMI, January 2008
 Vittorio Ferrari, Frederic Jurie, and Cordelia Schmid, Accurate Object Detection with Deformable Shape Models Learnt from Images, CVPR 2007, Minneapolis, USA
 Vittorio Ferrari, Frederic Jurie, and Cordelia Schmid, From Images to Shape Models for Object Detection, IJCV 2009 (to appear)