README.md

---
license: other
task_categories:
- object-detection
- image-segmentation
- robotics
- slam
- navigation
language:
- en
- es
tags:
- agriculture
- robotics
- orchards
- alto-valle
- stereo-vision
- lidar
- gnss-rtk
pretty_name: Alto Valle Dataset
size_categories:
- 100G<n<1T
viewer: false
---

# Alto Valle Dataset

<div align="center">
  <img src="https://raw.githubusercontent.com/Seba-san/AltoValleDataset/main/gifs/abril.gif" width="400"/>
  <p><em>Autonomous navigation data in Alto Valle Pear Orchards</em></p>
</div>

## 📖 Overview
The **Alto Valle Dataset** is a collection of experimental data acquired in pear orchards (*Pyrus communis*) at the **INTA Alto Valle Experimental Station** (Río Negro, Argentina).

The main goal of this dataset is to support the development of **localization, mapping (SLAM), and navigation algorithms** in agricultural environments. It presents unique challenges such as:
*   High variability in lighting and terrain.
*   Seasonal changes: Data captures during **Summer/Autumn** (leafy, pre-harvest) and **Winter** (leafless, pruning).
*   Repetitive structures (rows of trees).

## 🚜 Hardware & Setup
The data was captured using a modified **N. Blosi Senior** harvesting platform moving at approximately $0.3 m/s$.

| Sensor Type | Model | Specifications |
| :--- | :--- | :--- |
| **Stereo Camera** | **Stereolabs ZED** | 720x1280 @ 15 FPS. USB 3.0 connected to Nvidia Jetson TX1. |
| **LiDAR** | **SICK LMS-100** | 2D LiDAR, 50Hz (stored at 1Hz), 270° FOV, 0.5° resolution. |
| **GNSS-RTK** | **U-BLOX C94-M8P-2** | Base + Rover setup (915MHz link). ~2.5cm accuracy. 1Hz update rate. |

The sensors were mounted 2.5m high, with the camera acting as the reference frame.

## 📁 Dataset Structure
The dataset is organized by sequences (April and August). Each sequence contains:

```text
dataset/
├── sequenceXX/
│   ├── images/
│   │   ├── left_<index>.png    # Rectified left image
│   │   ├── right_<index>.png   # Rectified right image
│   │   └── timestamps.txt      # Image timestamps
│   ├── lidar.csv               # LiDAR readings
│   └── gnss.csv                # GNSS-RTK readings
```

### Data Format Details

*   **Images:** PNG format, rectified.
*   **GNSS (`gnss.csv`):**
    Format: `[latitude | longitude | timestamp]`
*   **LiDAR (`lidar.csv`):**
    Format: `[timestamp | nscan | 541 x range]`
    *   `nscan`: Frame number generated by the sensor.
    *   `range`: 541 distance values (0.5° angular resolution).

## 📐 Calibration Parameters
Intrinsic parameters for the **ZED Camera** used in this dataset. The baseline is **120.647 mm**.

| Parameter | Left Camera | Right Camera |
| :--- | :--- | :--- |
| **fx** | 692.964 | 698.848 |
| **fy** | 692.964 | 698.848 |
| **cx** | 576.186 | 737.995 |
| **cy** | 367.798 | 361.795 |
| **k1** | -0.182798 | -0.1634 |
| **k2** | 0.0277213 | 0.0214219 |

*Extrinsic parameters (transformations between sensors) are available in the PDF documentation or the paper.*

## 📅 Seasons & Conditions
The dataset covers different phenological stages of the pear crops (Williams, Abate Fetel, and Beurré D’Anjou varieties):

1.  **April 09, 2018 (Autumn):**
    *   **Condition:** Pre-harvest. Dense foliage, branches weighed down by fruit.
    *   **Weather:** Sunny, variable lighting (shadows/direct sun).
    *   **Ground:** Presence of weeds, irrigation ditches.
2.  **August 06, 2018 (Winter):**
    *   **Condition:** Pruning season. Deciduous trees (no leaves). Visible trunks and structure.
    *   **Weather:** Partly cloudy/overcast.
    *   **Ground:** Cleared weeds.

## 📥 Access & Download
The data is split into `.tar` archives due to size. You can download them directly from the **[Files and versions](https://huggingface.co/datasets/Seba-san/AltoValleDataset/tree/main)** tab.

## ⚖️ Legal Notice & Citation
The content of this database is under **Copyright** of the **Universidad Nacional del Comahue** and **INTA EEAV**.

If you use this dataset in your research, please cite the work presented at **Jornadas Argentinas de Robótica (JAR) 2022**:

*   **Paper:** [Read PDF (Spanish)](https://github.com/Seba-san/AltoValleDataset/blob/main/AVD_v0.pdf)
*   **Video:** [Watch on YouTube](https://youtu.be/qrSIFyLzFrQ)

```bibtex
@inproceedings{alto_valle_dataset_2022,
  title={Alto Valle Dataset: colección de datos experimentales enfocados en el estudio y desarrollo de algoritmos de navegación mediante visión en ambientes frutícolas},
  author={Sansoni, Sebastian and Raverta Capua, Francisco and Moreyra, Marcelo L. and Benitez Piccini, Edgardo},
  booktitle={Jornadas Argentinas de Robótica (JAR)},
  year={2022},
  organization={Universidad Nacional del Comahue & INTA EEAV},
  url={https://github.com/Seba-san/AltoValleDataset}
}
```

## Acknowledgments
This work was supported by the Project of Social Technological Development (PDTS) "Sistemas de Asistencia al Productor y Automatización de Máquinas para la Fruticultura de la Norpatagonia" (PDTS404), funded by CIN and CONICET.
```