PulseAugur
EN
LIVE 09:51:17

New methods improve depth estimation for fisheye cameras · 2 sources tracked

By PulseAugur Editorial · [2 sources] ·

Researchers have developed new methods to improve depth estimation from fisheye cameras. One approach, "Calibration Tokens," adapts existing foundational monocular depth estimators to fisheye images without retraining by aligning latent embeddings. Another method, OmniDS, uses a dual-stream context fusion technique with iterative refinement to handle visibility conflicts inherent in omnidirectional camera rigs. Both methods aim to enhance the accuracy and applicability of depth estimation for fisheye camera systems. AI

IMPACT Enhances the utility of fisheye cameras for applications requiring accurate 3D scene understanding.

RANK_REASON Two distinct research papers proposing novel methods for depth estimation from fisheye cameras.

Read on arXiv cs.AI →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New methods improve depth estimation for fisheye cameras · 2 sources tracked

COVERAGE [2]

  1. arXiv cs.AI TIER_1 English(EN) · Rit Gangopadhyay, Jung-Hee Kim, Xien Chen, Patrick Rim, Hyoungseob Park, Alex Wong ·

    Extending Foundational Monocular Depth Estimators to Fisheye Cameras with Calibration Tokens

    arXiv:2508.04928v5 Announce Type: replace-cross Abstract: We propose a method to extend foundational monocular depth estimators (FMDEs), trained on perspective images, to fisheye images. Despite being trained on tens of millions of images, FMDEs are susceptible to the covariate s…

  2. arXiv cs.CV TIER_1 English(EN) · Chaesong Park, Jihyeon Hwang, Muyeol Sung, Jongwoo Lim ·

    OmniDS: Dual-Stream Context Fusion for Omnidirectional Depth from Fisheye Cameras

    arXiv:2607.03038v1 Announce Type: new Abstract: Omnidirectional depth estimation from multi-fisheye camera rigs is complicated by visibility conflicts: wide baselines cause different cameras to observe different portions, or even different faces, of the same object, so aggregatin…

RELATED ENTITIES

RELATED TOPICS