The Plantas image database is composed by 9,398 images in the Plantas50Basic subset, 1,277 in the Plantas50Extra subset, and 22,661 images in the Plantas50Internet subset for 50 different species and cultivars. The Plantas50Basic and Plantas50Extra set have high-quality images taken from digital cameras and smartphones. They have a resolution of 2048x1536, and were taken at gardens and parks in Brazil during the months of December/2015 and March/2016.
You can use the python downloader that will automatically download the whole database to your computer. The total size is about 16GB and you should have at least 32GB free for the download and merge operations.
To save the dataset in the same directory of the script, run:
python download_plantas50.py
To save in another directory, run:
python download_plantas50.py /path/to/dir
You can also access: https://1drv.ms/f/s!AjZCiYkckpt_g-cafayn0qe-FyIr9g and download from there. Then you can run:
cat Plantas50.tar.part* > Plantas50.tar
or in Windows
copy /b Plantas50.tar.part* Plantas50.tar
and you'll be ready to go!
BY DOWNLOADING YOU ACCEPT TO USE THE INTERNET SUBSET FOR RESEARCH OR NON-COMMERCIAL USE ONLY. SEE LEGAL SECTION.
We finetuned a Mobile v2 model Plantas50 using Tensorflow and Keras as a simple example. You can check the notebook.
We also trained the Plantas50 in a Xception model in this notebook.
You can create TFRecord files by using our script:
python prepare_tfrecord_plantas50.py /path/to/Plantas50orpython prepare_tfrecord_plantas50.py /path/to/Plantas50 HeightxWidth
Supplemental Material for the paper 'Visual Recognition of Plant Species in the Wild' can be found in the paper-data folder.
| Label | Basic | Extra | Internet | Extended |
|---|---|---|---|---|
| Agave americana 'Marginata' | 201 | 0 | 240 | 441 |
| Agave angustifolia | 236 | 4 | 370 | 610 |
| Agave attenuata | 200 | 64 | 1036 | 1300 |
| Agave ovatifolia | 203 | 3 | 318 | 524 |
| Allamanda blanchetii | 201 | 0 | 380 | 581 |
| Allamanda cathartica | 200 | 4 | 1002 | 1206 |
| Alpinia purpurata | 201 | 0 | 1137 | 1338 |
| Anthurium andraeanum | 201 | 120 | 760 | 1081 |
| Beaucarnea recurvata | 191 | 11 | 763 | 965 |
| Begonia × hybrida | 100 | 0 | 226 | 326 |
| Bismarckia nobilis | 197 | 4 | 556 | 757 |
| Bougainvillea glabra | 196 | 4 | 726 | 926 |
| Buxus microphylla | 219 | 24 | 195 | 438 |
| Callistemon spp | 202 | 4 | 415 | 621 |
| Clerodendrum × speciosum | 201 | 63 | 174 | 438 |
| Codiaeum variegatum 'Aureo-maculatum' | 221 | 0 | 87 | 308 |
| Cordyline fruticosa | 201 | 1 | 551 | 753 |
| Cupressus sempervirens | 200 | 85 | 315 | 600 |
| Cycas revoluta | 221 | 91 | 833 | 1145 |
| Cycas thouarsii | 203 | 4 | 225 | 432 |
| Davallia fejeensis | 200 | 77 | 136 | 413 |
| Dianella ensifolia | 199 | 3 | 211 | 413 |
| Dieffenbachia amoena | 88 | 3 | 218 | 309 |
| Dracaena marginata | 115 | 88 | 248 | 451 |
| Duranta erecta 'Gold Mound' | 200 | 0 | 89 | 289 |
| Dypsis lutescens | 203 | 89 | 410 | 702 |
| Echeveria glauca | 200 | 114 | 290 | 604 |
| Eugenia sprengelii | 209 | 23 | 26 | 258 |
| Hibiscus rosa-sinensis | 211 | 3 | 1208 | 1422 |
| Impatiens hawkeri | 100 | 0 | 549 | 649 |
| Ixora coccinea | 200 | 32 | 617 | 849 |
| Ixora coccinea 'Compacta' | 201 | 117 | 215 | 533 |
| Justicia brandegeana | 214 | 0 | 600 | 814 |
| Leea guineensis | 100 | 5 | 59 | 164 |
| Loropetalum chinense | 204 | 2 | 729 | 935 |
| Monstera deliciosa | 220 | 0 | 806 | 1026 |
| Nematanthus wettsteinii | 191 | 48 | 97 | 336 |
| Nerium oleander | 195 | 12 | 1236 | 1443 |
| Ophiopogon jaburan | 210 | 6 | 123 | 339 |
| Philodendron imbe | 225 | 0 | 20 | 245 |
| Philodendron martianum | 97 | 3 | 120 | 220 |
| Phoenix roebelenii | 213 | 2 | 330 | 545 |
| Podocarpus macrophyllus | 98 | 2 | 374 | 474 |
| Rhapis excelsa | 201 | 75 | 687 | 963 |
| Rhododendron simsii | 218 | 0 | 485 | 703 |
| Russelia equisetiformis | 205 | 9 | 641 | 855 |
| Strelitzia reginae | 200 | 67 | 1075 | 1342 |
| Syngonium angustatum | 198 | 2 | 38 | 238 |
| Zamioculcas zamiifolia | 97 | 3 | 490 | 590 |
| Zinnia peruviana | 191 | 7 | 225 | 423 |
Plantas50Basic and Plantas50Extra by Rene Octavio Queiroz Dias are licensed under a Creative Commons Attribution 4.0 International License.
Some images of Plantas50Internet subset may have copyright. Training and using recognition model for research or non-commercial use may constitute fair use of data.
All code is under MIT license, unless stated otherwise in the header of the code. Or according to LICENSE or COPYING files inside the folders.
If the Plantas50 database was useful in your publications, please cite:
@inproceedings{plantas-db-2016,
author={Dias, Ren{\'e} Octavio Queiroz and Borges, D{\'i}bio Leandro},
booktitle={2016 IEEE International Symposium on Multimedia (ISM)},
title={Recognizing Plant Species in the Wild: Deep Learning Results and a New Database},
year={2016},
pages={197-202},
doi={10.1109/ISM.2016.0047},
isbn={978-1-5090-4571-6/16},
url={https://doi.org/10.1109/ISM.2016.0047},
month={Dec},}
If you are interested in how all these models work, you can check my Master Thesis.