Romea / cropcraft

CropCraft CropCraft is a python script that generates 3D models of crop fields, specialized in real-time simulation of robotics applications. • Designed for real-time simulation • Suitable for use with LiDARs and cameras • Highly configurable (YAML file) • Provide ground truth data (identify plant types in LiDAR data) Requirements This program uses blender as a backend. It is a 3D modeling software that you can dowload from the official website. If you use Ubuntu, you can install it using snap: The minimal required version is . Ensure that blender is launchable from the command line. It means that blender must be accessible using the environment variable. You also need to install some python requirements: Running To generate a crop field, you first need to create a configuration file (YAML formats). Some examples are available in the directory. Then you can execute the script and specify the path of the chosen configuration file. This command will generate a blender file named and a gazebo model named Some options are available and described using Image capture CropCraft can render a dataset of paired RGB images and semantic segmentation masks by adding a block to your configuration file: When a block is present, CropCraft renders two passes in a single run: • **RGB images** — rendered with Cycles, saved as JPEG in • **Semantic masks** — rendered with EEVEE using flat emission materials, saved as PNG in • Crops → (default: green ) • Weeds → (default: red ) • Ground / stones → (default: black ) > **Note:** when using , the desired GPU must be enabled in Blender > beforehand. Open Blender and go to **Edit > Preferences > System > Cycles Render Devices**, > then select your GPU. This setting persists across runs. (Verified on Blender 5.0.1.) See for a complete example. Documentation • Description of the configuration file format • How to add your own assets