<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    
    <title>hdddhdd.log</title>
    
    
    <description>This website is a virtual proof that I&apos;m awesome</description>
    
    <link>https://hdddhdd.github.io/</link>
    <atom:link href="https://hdddhdd.github.io/feed.xml" rel="self" type="application/rss+xml" />
    
    
      <item>
        <title>[논문 리뷰] DGD: Dynamic 3D Gaussians Distillation(ECCV 2024)</title>
        <description>
          Gaussian Splatting - 
          연구 배경 3D 장면 표현은 컴퓨터 비전에서 중요한 연구 분야이며, 최근 3D Gaussian Splatting(3DGS)이 실시간으로 고품질의 3D 렌더링을 가능하게 하는 기법으로 주목받고 있음. 기존 연구들은 3D 장면의 색상 및 형상을 잘 표현하지만, 시맨틱 정보(semantic information) 를 효과적으로 포함하지 못함. DGD는 3D Gaussian 표현을 기반으로 동적 3D 장면의 appearance, geometry, 그리고...
        </description>
        <pubDate>Sun, 09 Mar 2025 00:00:00 -0500</pubDate>
        <link>https://hdddhdd.github.io/2025-03-09-pr-dgd/</link>
        <guid isPermaLink="true">https://hdddhdd.github.io/2025-03-09-pr-dgd/</guid>
      </item>
    
      <item>
        <title>[코드 리뷰] OpenPCDet Docker로 실행하기</title>
        <description>
          3D Object Detection - 
          OpenPCDet 이란? OpenPCDet github에는 다음과 같이 나와있습니다! OpenPCDet is a clear, simple, self-contained open source project for LiDAR-based 3D object detection. OpenPCDet을 활용하면 KITTI, Waymo, nuScenes 등과 같이 Lidar-based 데이터셋에 대해 3D Object Detection을 수행할 때 기존 모델들을 쉽게 테스트 해볼 수 있습니당! OpenPCDet을 Docker에서 활용해보자! Docker 이미지 다운로드 sudo...
        </description>
        <pubDate>Sun, 02 Mar 2025 00:00:00 -0500</pubDate>
        <link>https://hdddhdd.github.io/2025-03-02-cd-openpcdet-docker/</link>
        <guid isPermaLink="true">https://hdddhdd.github.io/2025-03-02-cd-openpcdet-docker/</guid>
      </item>
    
      <item>
        <title>[논문 리뷰] Joint Homophily and Heterophily Relational Knowledge Distillation for Efficient and Compact 3D Object Detection (ACM MM 2024)</title>
        <description>
          Homophily · Heterophily · Relational Knowledge Distillation · 3D Object Detection - 
          논문 요약: 본 논문에서는 LiDAR 기반 3D 객체 탐지에서 동질성과 이질성을 고려한 새로운 지식 증류 방법인 Joint Homophily and Heterophily Relational Knowledge Distillation (H2RKD)을 제안함. 연구 배경 3D 객체 탐지 모델의 성능 향상을 위해 Knowledge Distillation(KD)이 적용되나, Point Cloud의 희소성, 불규칙성, 기하학적 복잡성으로 인해 KD의 일반화 능력이 저해됨. 기존 연구들은...
        </description>
        <pubDate>Sun, 23 Feb 2025 00:00:00 -0500</pubDate>
        <link>https://hdddhdd.github.io/2025-02-23-pr-h2rkd/</link>
        <guid isPermaLink="true">https://hdddhdd.github.io/2025-02-23-pr-h2rkd/</guid>
      </item>
    
      <item>
        <title>[논문 리뷰] Cross-Modality Knowledge Distillation Network for Monocular 3D Object Detection(ECCV 2022)</title>
        <description>
          Cross-Modality · Knowledge Distillation · Monocular 3D Object Detection - 
          논문 요약: 본 논문에서는 LiDAR 기반 탐지기의 지식을 단일 카메라(단안) 기반 탐지기로 효과적으로 전달하는 새로운 지식 증류 방법을 제안함. 연구 배경 3D 객체 탐지는 증강 현실, 로봇 공학, 자율주행 등 다양한 분야에서 핵심 기술로 활용되며, 단안 카메라 기반 탐지는 저비용·고해상도 데이터 획득이 가능하여 주목받고 있음. 그러나 단안 카메라 기반 3D...
        </description>
        <pubDate>Sun, 16 Feb 2025 00:00:00 -0500</pubDate>
        <link>https://hdddhdd.github.io/2025-02-16-pr-cmkd/</link>
        <guid isPermaLink="true">https://hdddhdd.github.io/2025-02-16-pr-cmkd/</guid>
      </item>
    
      <item>
        <title>[논문 리뷰] Good Teachers Explain: Explanation-Enhanced Knowledge Distillation(ECCV 2024)</title>
        <description>
          Model Compression · Faithful Distillation · Interpretability - 
          논문 요약: 본 논문에서는 Explanation-Enhanced Knowledge Distillation 방법을 제안하며, 이를 통해 학생 모델이 단순히 교사 모델의 출력을 모방하는 것이 아니라, 교사 모델이 학습한 설명의 유사성까지 유지하도록 학습할 수 있음을 보여줌. 연구 배경 Knowledge Distillation(KD)는 큰 교사 모델의 지식을 작은 학생 모델로 전달해서 경량 모델을 효과적으로 학습하는 기법이다. 하지만 기존 KD...
        </description>
        <pubDate>Tue, 11 Feb 2025 00:00:00 -0500</pubDate>
        <link>https://hdddhdd.github.io/2025-02-11-pr-e2kd/</link>
        <guid isPermaLink="true">https://hdddhdd.github.io/2025-02-11-pr-e2kd/</guid>
      </item>
    
  </channel>
</rss>
