Content-Based Image Retrieval: Plant Species Identification using Dynamic Programming


Isolated (hand picked) leaf	Digitized Herbarium Sample

Build and train a system using hand picked isolated leaves to predict plant-species of unknown isolated leaves
Build and train a system using isolated leaves to predict plant-species for a digitized herbarium samples
Build and train a system using herbarium samples to predict species of an unknown plant leaf

How

Using Dynamic Programming and concepts from
Dynamic Time Warping used for speech recognition
DNA (biological) sequence alignment algorithm

Results

Training Set	Test Set	Accuracy
Hand picked isolated leaves	Hand picked isolated leaves	96%
Hand picked isolated leaves	Herbarium Samples	59%
Herbarium Samples	Herbarium Samples	61%

Sample plant species and Code

Browse images of Herbarium samples

Content-Based Image Retrieval: Plant Species Identification (PDF)

Last Modified on September 27, 2002
contact: gandhias@cs.orst.edu