AgroScope PhENotyping tool
ASPEN project has as main goal to create a tool that can be used along diferent domains of agriculture (i.e. greenhouses, orchards, field, etc.) as a phenotyping tool for identify, quantify and measure diferent topics of agricultural interestes as for example yield, pest and diseases presence, stress, etc. For this, a group of diferent sensors are used to create a 3D reconstruction of the scanned environment what allow to obtain relevant data in two different configurations (Real time and postprocessing).
ASPEN project rely in several other works, here mentioned:
This repository is research in process!
Several improvements are in the pipeline of the project. Either contributions or personal work from the author can implement them in the near future
- Add depth to color matrix transformation for better alignments between object detection (2D) and point cloud (3D)
- Allow diferent objet detection algorithms
- Allow different tracking algorithms
- Implement YOLOv6
- Speed calculator for optimal high resolution cover
- Expose LiDAR intensity values
- Point cloud colorizing using multispectral values (radiance)
- Point cloud correction using DSL2 light sensor for radiance to reflectance transformation
- Implement SLAM loop closure using A-LOAM or similar
- GPS integration for point cloud alignment as postprocesing
- Tool transference to rover car for automatic data adquisition