ParticleTrieur is an open-source software providing integrated solutions for image preprocessing, annotation, labeling, metadata management, automated classification via Convolutional Neural Networks (CNNs), and morphometrical analyses.
ParticleTrieur is designed to help biologists and paleontologists build coherent datasets through an intuitive, non-supervised interface. It also allows users to easily develop custom CNNs that can be applied to new image datasets or integrated into automated systems. ParticleTrieur’s versatility makes it applicable to modern oceanographic research, paleoclimatic studies, and biostratigraphy.
https://github.com/microfossil/particle-trieur
- Image annotation
- Image preprocessing
- Image postprocessing
- CNN training
- Inference
- Morphometrics
Particle-Trieur calls the miso python library for all ML tasks.
Reference : Marchant, R., M. Tetard, A. Pratiwi, M. Adebayo, and T. de Garidel-Thoron. 2020. “Automated Analysis of Foraminifera Fossil Records by Image Classification Using a Convolutional Neural Network.” Journal of Micropaleontology. https://jm.copernicus.org/articles/39/183/2020/