Flavia (at a glance)
a Leaf Recognition Algorithm for Plant Classification using PNN (Probabilistic Neural Network)
Please cite our paper if you use our data and program in your publications. We will be very happy if you give us the credit.
This program is based on the paper A Leaf Recognition Algorithm for Plant classification Using Probabilistic Neural Network, by Stephen Gang Wu, Forrest Sheng Bao, Eric You Xu, Yu-Xuan Wang, Yi-Fan Chang and Qiao-Liang Xiang, published at IEEE 7th International Symposium on Signal Processing and Information Technology, Dec. 2007.
We also found some errors in the Latin or English names of plants. We listed the up-to-date version in the "Dataset" section. We shared our dataset for other researchers here.
The purpose of this MATLAB program is to teach a computer to classify plants via their leaves. You just need to input the leaf image of plant (acquired via digital camera or scanners), then the computer can tell you what kind of plant it is. Presently, our system can classify 32 plants. The average accuracy is 93% for all of them.
We utilize the PNN (Probabilistic Neural Network) to implement this AI process. 12 characters of leaves are taken into account, including geometrical ones and morphological ones. After discriminant analysis (stepwise method), all these characters are reserved. PCA orthogonalizes these 12 characters into 5 principal variables, which are input vectors of the PNN. Details can be found in our paper.
More details can be found from our SourceForge summary page.
The default download is our MATLAB source code. We strongly recommend you read the user manual first. If you need the leaf comparison function of our program, please also download the standard leaf image library and place it in proper path.
If any link is broken, please check all files of this project to determine. If you have further trouble, please ask questions in our mailing list or forum.
If you want a big collection of leaf images, please see the section below to download our 1GB dataset.
During our research, we suffered a lot from the lack of a standard plant leaf dataset. Thus, we don't have a benchmark to compare our algorithm with others. A public dataset may help other researchers working on similar projects as ours. So we decide to share our raw data. You can download the complete raw dataset. It is a very large file, around 1 GB. You can check your file integrity by this MD5SUM: 8d3ca661e201f4eac8d0975e7b6b5853.
When citing the source of the dataset, please don't use the web link, which might change. Please cite it as the data used in our paper: Stephen Gang Wu, Forrest Sheng Bao, Eric You Xu, Yu-Xuan Wang, Yi-Fan Chang and Chiao-Liang Shiang, A Leaf Recognition Algorithm for Plant classification Using Probabilistic Neural Network, IEEE 7th International Symposium on Signal Processing and Information Technology, Dec. 2007, Cario, Egypt
In our dataset, file names of all images are 4-digit numbers, followed by a ".jpg" suffix. The plants and their corresponding image file names are listed in below table. The classification labels of plants used in our program are listed at the most left column. Classification information from USDA websites, Wikipedia or other websites are listed in the most right column.
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. http://www.gnu.org/copyleft/gpl.html
Program information
Detailed information of our algorithm
Start the program from command window
Click to mark the two terminals of the longest and main vein of the leaf.
Side-by-side display of the image you inputed and the standard leaf image.
Result message box
Project file structure
Last update: Dec. 24, 2009