This repository presents results of our work to predict plant species based on the image of leaf using deep learning. The results are available in the Notebook
We implemented transfer learning using Alexnet with PyTorch. The following hyperparameters provided best accuracy of 91.7%.
Arch = 'alexnet'; Batch = 32; Hidden_units = 4096; Epochs = 200; Dropout = 0.5; Learning Rate = 0.01, Optmizer = SGD, Momentum = 0.9
Following Graphs show Model Accuracy for Training and Testing Phase; Model Loss for Training and Testing Phase; and Computation Time for Training the model and Training plus Tresting the model.
Our experimental dataset has following 11 types of plants.
| Folder ID | Plant Name |
|---|---|
| 1 | Acalypha hispida [EUPHORBIACAE] |
| 2 | Bauhinia coccinea [FABACEAE] |
| 3 | Calotropis gigantea [APOCYNACEAE] |
| 4 | Clitoria ternatea [FABACEAE] |
| 5 | Dillenia suffruticosa [DILLENIACEAE] |
| 6 | Ficus deltoidea [MORACEAE] |
| 7 | Melastoma beccarianum [MELASTOMATACEAE] |
| 8 | Melastoma malabathricum [MELASTOMATACEAE] |
| 9 | Melastoma malabathricum var alba [MELASTOMATACEAE] |
| 10 | Passiflora foetida [PASSIFLORACEAE] |
| 11 | Petrea volubilis [VERBENACEAE] |
We created dataset of 740 images from 11 plants and the dataset was divided into 596 images for training the model and 144 images used for testing the model.
Please submit your feedback to Nagender Aneja. Please write an email ([email protected]) if you are interested to impement the model in a mobile app or web app. We welcome people and organization who can provide more data on plants from different countries to join this project.
- Dr. Nagender Aneja, [email protected]
- Dr. Somnuk Phon-Amnuaisuk [email protected]
- Dr. Sandhya Aneja, [email protected]
- Dr. Hj Abd Ghani bin Hj Naim, [email protected]
- Dr. Rahayu Sukmaria binti Hj Sukri, [email protected]
- Rodzay bin Hj Abd Wahab [email protected]
- Nurul Hazlina binti Hj Zaini, [email protected]