Why-So-Deep: Towards Boosting Previously Trained Models for Visual Place Recognition

TL;DR

This paper introduces MAQBOOL, a method to enhance pre-trained models for visual place recognition, improving recall in SLAM systems with lower-dimensional descriptors and spatial information utilization.

Contribution

The paper proposes MAQBOOL, a novel approach that boosts existing pre-trained models for better image retrieval in SLAM without retraining or high-dimensional descriptors.

Findings

Achieves comparable results with 512-D descriptors to state-of-the-art 4096-D methods.

Uses spatial information to improve recall rate.

Effective in real-time multiagent SLAM systems.

Abstract

Deep learning-based image retrieval techniques for the loop closure detection demonstrate satisfactory performance. However, it is still challenging to achieve high-level performance based on previously trained models in different geographical regions. This paper addresses the problem of their deployment with simultaneous localization and mapping (SLAM) systems in the new environment. The general baseline approach uses additional information, such as GPS, sequential keyframes tracking, and re-training the whole environment to enhance the recall rate. We propose a novel approach for improving image retrieval based on previously trained models. We present an intelligent method, MAQBOOL, to amplify the power of pre-trained models for better image recall and its application to real-time multiagent SLAM systems. We achieve comparable image retrieval results at a low descriptor dimension (512-D), compared to the high descriptor dimension (4096-D) of state-of-the-art methods. We use spatial information to improve the recall rate in image retrieval on pre-trained models.