Overview of re_case
mainre_case is a utility crate within the rerun ecosystem used for performing case conversions according to the specific naming conventions required by Rerun.Rerun Documentation
repository·main·Indexed 27 days ago
https://github.com/rerun-io/rerunRerun is a multimodal data layer for physical AI designed to log, query, visualize, and stream complex sensor data such as images, point clouds, and transforms. It features a high-performance columnar storage format, a real-time viewer for debugging robotics and computer vision pipelines, and a comprehensive ecosystem of Rust crates for data handling, including gRPC communication via the Rerun Data Protocol (redap), Apache Arrow integration, and support for .rrd, HDF5, MCAP, and MP4 formats.
What's inside Rerun
Overview of re_query
mainre_queryis a crate within thererunfamily that provides high-level query APIs. It is designed to facilitate querying data within the Rerun ecosystem.Overview of re_lenses_core
mainre_lenses_coreis a crate within thererunecosystem that provides core lens definitions, builders, and composable transformations for Apache Arrow arrays. It is designed to enable transformations that convert one Arrow array type to another while preserving structural properties such as row counts and null handling.Overview of re_grpc_server
mainThere_grpc_servercrate provides a gRPC server implementation for the legacy StoreHub API within the Rerun ecosystem. It is used to facilitate communication via the StoreHub protocol.Overview of the Rerun server crate
mainThere_servercrate provides an in-memory, open-source implementation of the Rerun server. It is designed to support most of the same gRPC endpoints as the commercial Rerun Hub service, but operates entirely in-memory for maximum simplicity. It is primarily used for testing purposes.Overview of re_arrow_util
mainre_arrow_utilis a utility crate within thererunecosystem designed to provide helper functions for working with Apache Arrow data structures.Overview of re_recording_panel
mainThere_recording_panelcrate provides the user interface components for the recording panel within the Rerun ecosystem. It is part of the broaderrerunfamily of crates used for visualizing and managing recorded data.Overview of the Rerun Viewer
mainThe Rerun Viewer is the primary crate responsible for the Rerun graphical user interface (GUI). It serves as the visual component of the Rerun ecosystem, allowing users to interact with logged data. It communicates with the Rerun server using gRPC via there_redap_clientcrate.Overview of Rerun's Data Layer and Catalog Server
mainRerun acts as a Unified Data Layer for Physical AI. The Rerun SDK connects to a catalog server to enable storing, retrieving, and querying large datasets. This integration allows for visual browsing and inspection of data through the Rerun viewer.
For local testing and experimentation, Rerun provides a simplified open-source catalog server that is API-compatible with the managed Rerun Hub offering. Note that the open-source server loads all data into memory, making it highly performant for small datasets but limited by available RAM.
Overview of Rerun components
mainRerun provides a unified data layer for multi-rate, multimodal robotics data, covering the journey from raw recordings to model training. It consists of two primary components:
- Rerun SDK: An open-source library and set of tools used for logging, storing, querying, visualizing, and training on multimodal data.
- Rerun Hub: A data catalog and backend designed for large-scale storage, access, and streaming of robotics data directly from object storage.
Overview of re_sorbet
mainre_sorbetis a crate within thererunecosystem used for managing Arrow metadata and record batch definitions. It is specifically designed to handle the structure of Arrow record batches and their associated metadata for various Rerun use cases.Overview of re_lenses
mainLenses is an API within thererunecosystem designed for extracting, transforming, and restructuring component data. It allows developers to manipulate how data is presented or accessed from the underlying storage.