Selectivity Drives Efficiency: Dataset Pruning for Visual Place Recognition
Computer Vision and Pattern Recognition
Summary
The authors address the problem of training visual place recognition (VPR) models with very large datasets, which can be expensive and slow. They propose a new method that prunes, or reduces, the dataset by focusing on entire places rather than individual images. Their approach evaluates places based on how diverse images within the place are and how similar the place is to others, keeping only the most informative places. Experiments show their method preserves model performance while cutting down on storage and training time.
Authors
Tong Jin, Yunpeng Liu, Shuyu Hu, Chun Yuan, Song Wang, Feng Lu
Abstract
Recent visual place recognition (VPR) studies have increasingly relied on large-scale datasets to train more robust and discriminative models. Although this trend significantly improves recognition performance, it also introduces substantial storage and training costs, especially when new architectures or training strategies need to be repeatedly developed and evaluated. Dataset pruning (DP) provides a promising way to improve data efficiency by retaining only informative training data. However, conventional DP methods mainly follow the sample-wise classification paradigm, which overlooks the relation-dependent training nature of VPR, where supervision is typically formed by image pairs rather than independent images. To address this issue, we propose a place-wise dataset pruning framework tailored for VPR. Instead of pruning individual images, our method treats each place as the basic pruning unit and introduces two complementary novel metrics, i.e., intra-place diversity (IPD) and inter-place similarity (IPS), to evaluate the training value of each place. By jointly considering these two metrics, our method ranks all places and constructs a compact yet informative coreset, thereby allowing the pruned dataset to still support the training of robust and discriminative VPR models. Extensive experiments demonstrate that our method consistently outperforms state-of-the-art DP baselines under different pruning ratios while reducing selection and training costs. Moreover, by pruning a merged dataset roughly 3.5$\times$ the size of GSV-Cities to a comparable scale, our coreset maintains highly competitive performance, achieving 94.5\% R@1 on MSLS-val and 97.0\% R@1 on Nordland with only NetVLAD. Codes will be made publicly available.