Geometry-Aware Fisheye-LiDAR Fusion for Robust 3D Object Detection in Low-Overlap Setups
Xiangzhong Liu, Xihao Wang, Hao Shen
実装難易度
Hard
推論・学習コスト
High
想定用途
検出
概要
As autonomous systems expand from capital-intensive robotaxis to cost-sensitive logistics, sensor configurations are increasingly optimized for coverage-per-cost. A prevalent sparse-view setup utilizes dual-fisheye cameras with a roof-mounted LiDAR, introducing severe geometric challenges: extreme radial distortion, minimal overlap, and misalignment between spherical projections and rectilinear…
何が新しいか
As autonomous systems expand from capital-intensive robotaxis to cost-sensitive logistics, sensor configurations are increasingly optimized for coverage-per-cost. A prevalent sparse-view setup…
何に使えるか
検出
実装情報
- Paper URL
- あり
実装チェックリスト
実装または配布ページ
要確認Paper onlyの可能性があるため再実装前提で確認してください。
一次情報リンク
OKPaper
検証しやすさ
要確認公式実装が見つからないため、論文から再実装する前提です。
計算資源
要確認学習や高解像度推論ではGPUメモリと実行時間に注意が必要です。
ライセンス
未取得配布元のLICENSE、モデルカード、Paperの利用条件を確認してください。
商用利用
未取得研究利用限定、データセット由来制限、API規約の有無を確認してください。
自社データで試すなら
製造業・材料開発のExcel/CSVデータに落とし込むための最初の手順です。
- 1まず自社データを、入力条件、目的変数、評価したい指標に分けて整理します。
- 2正常データだけで動くベースラインを作り、異常スコアのしきい値を現場知見と合わせます。
- 3評価指標はR2/RMSE、AUC、異常検知の再現率、実験回数削減率など、現場の意思決定に近いものを選びます。
- 4SHAPや特徴量重要度で、効いている因子が物理・化学・工程知識と矛盾しないか確認します。
実装難易度
Hard - 公式実装が見つからないため、論文から再実装する前提です。
必要リソース
- GPU目安: High
- データセット: 論文・リポジトリ側の指定を確認してください。
- 学習要否: 再学習や評価環境の準備が必要になる可能性があります。
- 学習や高解像度推論ではGPUメモリと実行時間に注意が必要です。
実務で使う場合の注意点
- ライセンスと商用利用条件は、Paper / GitHub / Hugging Face の配布元で確認してください。
- 精度、再現性、計算コストはデータセットや評価条件に依存します。
- 個人情報や機密データを扱う場合は、入力データの保存先と外部API利用条件を確認してください。
関連記事
Distortion-Aware PETR for BEV Object Detection with Mixed Pinhole-Fisheye Cameras
Fisheye cameras are widely deployed in autonomous driving perception suites for their low cost and full-covera
PAI: Preserving Amplitude Information in Representation-Based Time-Series Anomaly Detection
時系列データの異常の検出におけるアンプリットードの保存は、パフォーマンスを向上させるために不可欠です。
Visual Prompting Meets Feature Reconstruction-Based Anomaly Detection with Dual-Teacher Supervision
Recent Anomaly Detection methods achieve perfect detection and segmentation scores on well-established dataset
Securing Self-supervised Data Curation for Foundation Models Robustness
Self-supervised data curation provides a pathway to scaling and improving the generalization capabilities of m