Overview

We have created a 17 category flower dataset with 80 images for each class. The flowers chosen are some common flowers in the UK. The images have large scale, pose and light variations and there are also classes with large varations of images within the class and close similarity to other classes. The categories can be seen in the figure below. We randomly split the dataset into 3 different training, validation and test sets. A subset of the images have been groundtruth labelled for segmentation.

Instruction

This set contains images of flowers belonging to 17 different categories. The images were acquired by searching the web and taking pictures. There are 80 images for each category.

The datasplits are specified in datasplits.mat

There are 3 separate splits. The results in the paper are averaged over the 3 splits. Each split has a training file (trn1,trn2,trn3), a validation file (val1, val2, val3) and a testfile (tst1, tst2 or tst3).

Segmentation Ground Truth

The ground truth is given for a subset of the images from 13 different categories. More details can be found in the paper

Distance matrices

We provide two set of distance matrices:

distancematrices17gcfeat06.mat
distancematrices17itfeat08.mat

More details can be found in: Delving into the whorl of flower segmentation.

Citation

Please use the following citation when referencing the dataset:

@InProceedings{Nilsback06,
  author       = "Maria-Elena Nilsback and Andrew Zisserman",
  title        = "A Visual Vocabulary for Flower Classification",
  booktitle    = "IEEE Conference on Computer Vision and Pattern Recognition",
  volume       = "2",
  pages        = "1447--1454",
  year         = "2006",
}