Cauliflower centre detection and 3-dimensional tracking for robotic intrarow weeding

<h1>Cauliflower Center Dataset</h1>
<p>The dataset is used for autonomous intrarow weeding as described in the paper of Willekens et al. (2025).</p>

<blockquote>
Willekens, A., Callens, B., Pieters, J., Wyffels, F., & Cool, S. (2025).<br>
<i>Cauliflower centre detection and 3-dimensional tracking for robotic intrarow weeding. Precision Agriculture. Springer.
</blockquote>

<h2>Center</h2>
<p>The cauliflower centers in the images (PNG files) are provided in JSON files with the same name as the corresponding image. Each JSON file follows this structure:</p>

<pre><code class="json">
{
"centers": [
[
<x pixel 1>,
<y pixel 1>
],
[
<x pixel 2>,
<y pixel 2>
],
...
]
}
</code></pre>

<p>The <code>.pkl</code> files contain the same information as the JSON files but in a binary format.</p>

<h2>Segmentation</h2>
<p>Many of the images are also segmented. The segmentation masks are stored in files with the <code>.sam.png</code> extension.</p>

<h2>Dataset Overview</h2>
<p>The dataset file <code>db_dump.sql</code> provides the content of a PostgreSQL database that was used in the study of Willekens et al. (2025). When recovering this dataset, the PostGIS extension must be installed.</p>

<ol>
<li>Create a database in PostgreSQL <code>CREATE DATABASE cauliflower_db;</code></li>
<li>Connect to the new database and install the PostGIS plugin <code>CREATE EXTENSION postgis;</code>. See the
<a href="https://postgis.net/documentation/getting_started/">PostGIS installation instructions</a> if PostGIS is not yet installed.</li>
<li>Recover the database <code>psql -U your_username -d cauliflower_db -f db_dump.sql</code></li>
</ol>

<p>The tables contain the following information.</p>

<ul>
<li>
<strong>dataset</strong><br>
The datasets correspond to the different folders (e.g., <code>20230809.zip</code>, <code>20230810.zip</code>, <code>20230811.zip</code>), with the folder name representing the date of data collection (<code>YYYYMMDD</code>).
</li>
<li>
<strong>calibration</strong><br>
Refers to the calibration files stored in <code>calib.zip</code>.
</li>
<li>
<strong>image</strong><br>
Holds metadata for individual images, such as geometry, time, and the associated dataset.
</li>
<li>
<strong>label</strong><br>
Stores the references of the labels for each image.
</li>
<li>
<strong>label_set</strong><br>
Bundles different labels into one <code>label_set</code> used for training, validation, or testing.
</li>
<li>
<strong>label_set_entry</strong><br>
Serves as a junction table associating labels with label sets.
</li>
<li>
<strong>sensor</strong><br>
Lists the sensors used during data collection.
</li>
<li>
<strong>training</strong><br>
Describes the training sessions that were initiated. It refers to a validation and training <code>label_set</code> and specifies the training hyperparameters.
</li>
</ul>