PK PyTorchKR Papers
Papers with Code · CC BY-SA 4.0

3D Point Cloud Interpolation

200 평가 기록 27 datasets

Point cloud interpolation is a fundamental problem for 3D computer vision. Given a low temporal resolution (frame rate) point cloud sequence, the target of interpolation is to generate a smooth point cloud sequence with high temporal resolution (frame rate).

개요는 각 항목의 일부입니다. 위 탭에서 전체 기록을 페이지별로 탐색할 수 있습니다.

벤치마크

Papers with Code 아카이브의 데이터셋별 기록 예시입니다. 평가 조건과 지표 방향이 검증되지 않아 순위를 의미하지 않습니다.

Dataset기록된 모델 예시MetricPaperCode
GeDi
Feature Matching Recall: 0.982Recall (30cm, 5 degrees): 86.54
논문 코드
IMFNet
Feature Matching Recall: 98.6
논문 코드
NgeNet
Recall (0.3m, 15 degrees): 95.0RE (all): 4.932TE (all): 0.155
논문 코드
GeDi
Success Rate: 98.92
논문 코드
CoFF
Recall ( correspondence RMSE below 0.2): 81.6
논문 코드
NgeNet
Recall ( correspondence RMSE below 0.2): 92.9
논문 코드
GeoTransformer
Recall (0.6m, 5 degrees): 99.5
논문 코드
Greedy Grid Search
Recall (3cm, 10 degrees): 92.81RRE (degrees): 0.014RTE (cm): 0.009
논문 코드
FPFH+SC2-PCR
Recall (3cm, 10 degrees): 99.88RRE (degrees): 0.91RTE (cm): 0.43
논문 -
FPFH+SC2-PCR
Recall (3cm, 10 degrees): 38.85RRE (degrees): 2.42RTE (cm): 1.23
논문 -
Exhaustive Grid Search
Recall (3cm, 10 degrees): 88.06RRE (degrees): 0.030RTE (cm): 0.017
논문 코드
GeDi
Recall (3cm, 10 degrees): 99.76RRE (degrees): 1.629RTE (cm): 1.162
논문 코드

데이터셋

27개
ETH (trained on 3DMatch)3DMatch Benchmark3DMatch (at least 30% overlapped - FCGF setting)KITTI (trained on 3DMatch)3DLoMatch (10-30% overlap)3DMatch (at least 30% overlapped - sample 5k interest points)KITTI (FCGF setting)FPv1FP-O-EFP-O-HFP-O-MFP-R-E

논문

라이브러리

Papers with Code 리더보드 row에 연결된 구현 저장소입니다.

github.com/ispc-lab/neuralpci 1 논문
github.com/fabiopoiesi/gedi 1 논문
github.com/QingyongHu/SpinNet 1 논문
github.com/XuyangBai/D3Feat 1 논문
github.com/XuyangBai/D3Feat.pytorch 1 논문

하위 태스크

PC Point Cloud Registration 25 데이터셋 · 191 rows