Benchmark Analysis of YOLO Deep Learning Models for UAV-Based Bridge Inspection
This research benchmarks four YOLO deep learning model variants (YOLOv5, YOLOv6, YOLOv7, YOLOv8) for bridge component detection using the COCO-Bridge-2021+ dataset, identifying YOLOv8n, YOLOv7tiny, YOLOv6m, and YOLOv6m6 as offering the best balance between accuracy and processing speed for potential use in UAV-based bridge inspection systems.
FreeCap：開放環境中基於混合無需校準的多人動作捕捉
FreeCap 是一種新穎的混合無需校準方法，它結合了單個 LiDAR 和可擴展移動相機的優勢，能夠在開放環境中準確捕捉全局多人運動。
개방 환경에서의 하이브리드 캘리브레이션 없는 모션 캡처: FreeCap
단일 LiDAR와 확장 가능한 이동 카메라를 결합한 새로운 하이브리드 캘리브레이션 없는 모션 캡처 시스템인 FreeCap을 통해 개방 환경에서 여러 사람의 움직임을 정확하게 포착할 수 있습니다.
オープン環境におけるキャリブレーションフリーのハイブリッドモーションキャプチャ：FreeCap
本稿では、単一のLiDARと拡張可能な移動カメラを組み合わせた、オープン環境においてキャリブレーションフリーで正確なマルチパーソンモーションキャプチャを実現する、FreeCapと呼ばれる新しいハイブリッド手法を提案する。
Calibration-Free Multi-Person Motion Capture in Open Environments Using LiDAR and Moving Cameras: Introducing FreeCap
FreeCap is a novel hybrid system that leverages a single LiDAR and multiple moving cameras to achieve accurate and calibration-free multi-person motion capture in open environments.
一種高效的單幅像非均勻校正演算法
本論文提出了一種名為 MIRE 的演算法，用於校正非製冷紅外線影像中的非均勻性問題，該方法基於圖像直方圖均衡化技術，具有高效、即時且無需場景資訊等優點。
효율적인 단일 적외선 이미지 불균일 보정 알고리즘
본 논문에서는 냉각되지 않은 적외선 이미지에서 흔히 발생하는 불균일성(NU)을 보정하는 새롭고 효율적인 단일 이미지 알고리즘인 MIRE(Midway Infrared Equalization)를 제안합니다.
単一画像における非一様性の効率的な補正アルゴリズム
冷却型および非冷却型の赤外線画像に頻繁に見られる固定パターンノイズ（FPN）は、センサーの応答のばらつきによって発生し、画像品質を著しく低下させる。本稿では、単一画像からこのFPNを効率的に補正する新しいアルゴリズム「MIRE（Midway Infrared Equalization）」を提案する。
Efficient Single-Image Non-Uniformity Correction Algorithm for Uncooled Infrared Images: The Midway Infrared Equalization (MIRE) Method
This paper introduces a novel, real-time, single-image algorithm called Midway Infrared Equalization (MIRE) for correcting non-uniformity in uncooled infrared images, achieving superior performance compared to existing methods by leveraging the inherent information within individual image columns and a sliding midway histogram equalization technique.
圖像理解模型可用於構建高效的圖像生成分詞器
具備圖像理解能力的模型，例如透過知識蒸餾訓練的VQ-KD分詞器，在圖像生成任務中表現更出色，超越了傳統基於像素重建的VQGAN方法。
© 2024 by Linnk AI