FiLark: a streaming-first software framework for end-to-end exploration, annotation, and algorithm integration in distributed acoustic sensing
Researchers have developed FiLark, a new Python framework designed for distributed acoustic sensing (DAS) data. This framework adopts a streaming-first approach, enabling continuous exploration, annotation, and integration of algorithms with DAS data streams. FiLark supports interactive visualization of long recordings with constant memory usage and allows for direct event labeling within streams to create machine-learning-ready datasets. It also includes GPU-accelerated signal processing and a standardized interface for integrating real-time detectors and models. AI
IMPACT Enables more efficient processing and machine learning on continuous, high-volume sensor data streams.