This repository aims to provide a neural network that is capable of classifying plant species based on images of their leaves.
NOTE: Each leaf image is of some dimension L x W, where L = W or L != W. By slicing each images into rows, and appending each row to the one above it we form a 1 x (L * W) matrix to pass into the Input layer of our neural network. For example a 10 x 10 > pixel leaf image will be converted to a 1 x 100 matrix.
Layer | Size |
---|---|
Input | L * W |
Hidden-1 | [(L * W) + (C)] / 2 |
Output | C |
Some helpful notation for describing the neural network architecture design:
Symbol | Description |
---|---|
L | length of a leaf image training example |
W | width of a leaf image training example |
C | number of labels or classes for neural network |
We'll use open and free datasets to train our learning algorithms.
So far we have tested on the following datasets:
Name | Info | Download |
---|---|---|
One-hundred plant species leaves data set Data Set | UCI Machine Learning Repository | 100 leaves plant species.zip |
Leaf Data Set | UCI Machine Learning Repository | leaf.zip |