Unsupervised Metric Relocalization Using Transform Consistency Loss
This addresses the challenge of robust metric relocalization in dynamic environments for robotics and computer vision applications, offering a novel self-supervised alternative to supervised methods.
The paper tackles the problem of metric relocalization without requiring accurate image correspondences by proposing a self-supervised approach using a transform consistency loss. The result shows that their method outperforms supervised methods when limited ground-truth data is available, as demonstrated on synthetic and real-world datasets.
Training networks to perform metric relocalization traditionally requires accurate image correspondences. In practice, these are obtained by restricting domain coverage, employing additional sensors, or capturing large multi-view datasets. We instead propose a self-supervised solution, which exploits a key insight: localizing a query image within a map should yield the same absolute pose, regardless of the reference image used for registration. Guided by this intuition, we derive a novel transform consistency loss. Using this loss function, we train a deep neural network to infer dense feature and saliency maps to perform robust metric relocalization in dynamic environments. We evaluate our framework on synthetic and real-world data, showing our approach outperforms other supervised methods when a limited amount of ground-truth information is available.