Plant Species Identification

# Plant Species Identification

## Overview

This dataset contains a collection of annotated images featuring various plant species commonly found in a natural or cultivated environment.
The annotations consist of **bounding boxes (BBs)** marking the location of individual plants categorized into four main classes.

This dataset is ideal for training and evaluating object detection models in agricultural or ecological research contexts, specifically for tasks such as automated plant counting, localization, or preliminary species classification based on visual features.

## Data Structure

The dataset comprises **741 unique images** and is provided with annotations in the COCO JSON format.

### Plant Categories

The annotations classify plants into four main groups:

| Category ID | English Name | Portuguese Name | Description |
| :--- | :--- | :--- | :--- |
| 1 | **shrubs** | arbustos | Woody plants, smaller than a tree, typically having multiple main stems. |
| 2 | **grasses** | gramineas | Herbaceous, non-woody plants with narrow leaves growing from the base (e.g., cereals, pasture grasses). |
| 3 | **legumes** | leguminosas | Plants that produce seeds in pods, often having nitrogen-fixing properties. |
| 4 | **others** | outras | Any other plant type not covered by the preceding categories. |

---

## Dataset Statistics

The current release contains a total of **1744 bounding box annotations** (plants).
The distribution across the four categories is well-balanced:

### Summary Metrics

* **Total Number of Images:** 741
* **Total Number of Plants Annotated:** 1744

### Samples per Category

| Class Name (English) | Number of Samples | Percentage of Total |
| :--- | :--- | :--- |
| **shrubs** | 407 | 23.34% |
| **grasses** | 486 | 27.87% |
| **legumes** | 473 | 27.12% |
| **others** | 378 | 21.67% |
| **Total** | 1744 | 100.00% |