GRM: Gradient Rectification Module for Visual Place Retrieval
This work addresses a specific bottleneck in visual place retrieval for applications like robotics and navigation, but it is incremental as it builds on existing methods by adding a module to improve descriptor distribution.
The paper tackles the problem of visual place retrieval by addressing the issue of global descriptors falling into a low-dimensional principal space, which harms retrieval performance, and proposes a Gradient Rectification Module (GRM) that rectifies gradients to encourage more uniform descriptor generation, achieving improved results on multiple datasets.
Visual place retrieval aims to search images in the database that depict similar places as the query image. However, global descriptors encoded by the network usually fall into a low dimensional principal space, which is harmful to the retrieval performance. We first analyze the cause of this phenomenon, pointing out that it is due to degraded distribution of the gradients of descriptors. Then, we propose Gradient Rectification Module(GRM) to alleviate this issue. GRM is appended after the final pooling layer and can rectify gradients to the complementary space of the principal space. With GRM, the network is encouraged to generate descriptors more uniformly in the whole space. At last, we conduct experiments on multiple datasets and generalize our method to classification task under prototype learning framework.