FishRoPE: Projective Rotary Position Embeddings for Omnidirectional Visual Perception

TL;DR

FishRoPE introduces a novel spherical coordinate-based attention mechanism and adaptation framework that enables vision models to effectively handle fisheye camera distortions without retraining from scratch.

Contribution

The paper proposes FishRoPE, a lightweight, architecture-agnostic method that adapts frozen vision models to fisheye geometry using spherical attention reparameterization.

Findings

Achieves state-of-the-art results on WoodScape 2D detection with 54.3 mAP.

Attains 65.1 mIoU on SynWoodScapes BEV segmentation.

Introduces negligible computational overhead with FishRoPE.

Abstract

Vision foundation models (VFMs) and Bird's Eye View (BEV) representation have advanced visual perception substantially, yet their internal spatial representations assume the rectilinear geometry of pinhole cameras. Fisheye cameras, widely deployed on production autonomous vehicles for their surround-view coverage, exhibit severe radial distortion that renders these representations geometrically inconsistent. At the same time, the scarcity of large-scale fisheye annotations makes retraining foundation models from scratch impractical. We present \ours, a lightweight framework that adapts frozen VFMs to fisheye geometry through two components: a frozen DINOv2 backbone with Low-Rank Adaptation (LoRA) that transfers rich self-supervised features to fisheye without task-specific pretraining, and Fisheye Rotary Position Embedding (FishRoPE), which reparameterizes the attention mechanism in the spherical coordinates of the fisheye projection so that both self-attention and cross-attention operate on angular separation rather than pixel distance. FishRoPE is architecture-agnostic, introduces negligible computational overhead, and naturally reduces to the standard formulation under pinhole geometry. We evaluate \ours on WoodScape 2D detection (54.3 mAP) and SynWoodScapes BEV segmentation (65.1 mIoU), where it achieves state-of-the-art results on both benchmarks.