Planetary feature fields compress earth data with speed and accuracy
Planetary Feature Fields are Scalable Earth Representations
Computer Vision and Pattern RecognitionMachine Learning
Summary
Satellite data about Earth comes from many sources and is very large, making it hard to store and use together. The authors present Planetary Feature Fields (PFFs), a new way to combine these data into a compact, continuous representation that keeps important detail over time and space. PFFs compress data by over a thousand times while still performing well on tasks like recognizing land features or changes. They also allow adding new data easily and let users access information much faster than existing methods.
What this means in practice
- •For earth observation teams: Store multiple satellite data products in a compressed, unified form that supports fast and accurate access for analysis over time.
- •For environmental monitoring groups: Quickly detect and classify changes on Earth's surface using compact representations that maintain accuracy despite high compression.
Authors
Arjun Rao, Sebastian Loeschcke, Anthony Fuller, Isaac Corley, Nico Lang, Evan Shelhamer
Abstract
Satellite observations, precomputed embeddings, and map products describe the same evolving Earth, yet are stored as independent, petabyte-scale data products. Their continued growth calls for compact representations of multiple products while preserving spatial and temporal detail. We introduce Planetary Feature Fields (PFFs), which exploit redundancy across data products by modeling them jointly as continuous functions of space and time at planetary scale. PFFs are spatially local explicit-implicit (hybrid) neural fields. Each field shares a factored feature volume---a decomposition of an explicit 3D grid with smaller factors---across products, while lightweight implicit decoders reconstruct individual products across multiple timesteps. PFFs reconstruct EO products over space and time more accurately than single-product fields at matched compression rates. At $1800\times$ compression relative to the uncompressed source data, reconstructed features retain approximately $90\%$ or more of the performance achieved with the original features on pixel-level segmentation, change detection, and patch-level classification tasks. PFFs can add new timesteps by extending their factored feature volumes and add new products by attaching new decoders, while leaving existing outputs unchanged. PFFs reduce end-to-end feature access latency by an order of magnitude relative to evaluated API and cloud-storage pipelines.