Spectral Geometric Verification: Re-Ranking Point Cloud Retrieval for Metric Localization

Vidanapathirana, Kavisha; Moghadam, Peyman; Sridharan, Sridha; Fookes, Clinton

doi:10.1109/lra.2023.3255560

Cited by 12 publications

(10 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As for the global descriptors only, a widely used approach is Query Expansion (QE). In particular, αQE has been used both in image retrieval [12] and retrieval-based place recognition [13]. On the other hand, the category of methods that use both local and global information is more common in applications with geometrical data.…”

Section: B Re-rankingmentioning

confidence: 99%

“…These methods, known as geometrical verification (GV), work well when seeking false positives but are usually computationally demanding when working with large point clouds. More recently, a more efficient approach was proposed in [13] called spectral geometrical verification (SGV), which relies on correspondence compatibility graphs [21] to assess the spatial consistency of two point clouds.…”

Section: B Re-rankingmentioning

confidence: 99%

“…This section describes the proposed transformer-based re-ranking approach, TReR, depicted in Figure 2. Instead of following traditional point cloud-based loop re-ranking approaches, which rely predominantly on geometrical cues present in the scene [13], we adopt a more general information re-ranking approach, mainly used in image-related ranking tasks [3]. We are furthermore inspired by other works in context ranking, such as [23].…”

Section: B Transformer-based Re-rankingmentioning

confidence: 99%

“…Firstly, we give the results for the baseline methods without reranking and with re-ranking, using TReR (our approach) and αQE [12]. Secondly, we compare TReR to the spectral geometrical verification (SGV) method [13] and αQE, using as place recognition model LoGGNet3D-Net.…”

Section: Re-ranking Evaluationmentioning

confidence: 99%

“…TReR is evaluated on five different place recognition models [9], [10], [11], which serve as baselines. Additionally, we also compare TReR to another global descriptor-based re-ranking approach (αQE [12]) and a geometrical verification (GV) method (SGV [13]). The results are obtained on the KITTI dataset and indicate that TReR achieves a greater performance than the baselines (with no re-ranking) and αQE while offering an acceptable trade-off between performance and efficiency in contrast to SGV [13].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

AttDLNet: Attention-Based Deep Network for 3D LiDAR Place Recognition

Barros

Garrote

Pereira

et al. 2022

Lecture Notes in Networks and Systems

View full text Add to dashboard Cite

Robust and reliable place recognition and loop closure detection in agricultural environments is still an open problem. In particular, orchards are a difficult case study due to structural similarity across the entire field. In this work, we address the place recognition problem in orchards resorting to 3D LiDAR data, which is considered a key modality for robustness. Hence, we propose ORCHNet, a deep-learning-based approach that maps 3D-LiDAR scans to global descriptors. Specifically, this work proposes a new global feature aggregation approach, which fuses multiple aggregation methods into a robust global descriptor. ORCHNet is evaluated on real-world data collected in orchards, comprising data from the summer and autumn seasons. To assess the robustness, We compare ORCHNet with state-ofthe-art aggregation approaches on data from the same season and across seasons. Moreover, we additionally evaluate the proposed approach as part of a localization framework, where ORCHNet is used as a loop closure detector. The empirical results indicate that, on the place recognition task, ORCHNet outperforms the remaining approaches, and is also more robust across seasons. As for the localization, the edge cases where the path goes through the trees are solved when integrating ORCHNet as a loop detector, showing the potential applicability of the proposed approach in this task. The code and dataset will be publicly available at:https://github.com/Cybonic/ORCHNet.git

show abstract

Section: B Re-rankingmentioning

confidence: 99%

Section: B Re-rankingmentioning

confidence: 99%