Data Labeling

Data labeling encodes expert knowledge into harmonized, multimodal datasets, transforming raw signals into structured corpora that enable reliable machine learning, physics validation, and operational decision-making. The dFL GUI supports the following labeling approaches:

---

1. Manual Labeling

Manual Labeling allows users to custom label multimodal segments across records, setting their own labels types, properties, label colors, and label locations.

---

2. Autolabeling

Automated labeling leverages physics-informed algorithms, statistical methods, machine learning, and simulation-driven pipelines to efficiently generate scalable, consistent, and scientifically meaningful annotations from multimodal datasets.

---

3. Custom Autolabeling

Custom autolabelers can be seamlessly integrated into the dFL GUI through straightforward Python scripting.

---