Supervised Fine-tuning Evaluation for Long-term Visual Place Recognition

TL;DR

This study evaluates the effectiveness of fine-tuned deep CNNs with different pooling layers and loss functions for long-term visual place recognition under challenging conditions, demonstrating ArcFace loss as the most effective.

Contribution

It provides a comprehensive comparison of deep CNN architectures with various loss functions for visual place recognition, highlighting the superiority of ArcFace loss in end-to-end fine-tuning.

Findings

ArcFace loss improves recognition accuracy by 1-4% in outdoor datasets.

Fine-tuning with ArcFace outperforms triplet and contrastive losses.

Deep CNN features are effective for place recognition across indoor and outdoor environments.

Abstract

In this paper, we present a comprehensive study on the utility of deep convolutional neural networks with two state-of-the-art pooling layers which are placed after convolutional layers and fine-tuned in an end-to-end manner for visual place recognition task in challenging conditions, including seasonal and illumination variations. We compared extensively the performance of deep learned global features with three different loss functions, e.g. triplet, contrastive and ArcFace, for learning the parameters of the architectures in terms of fraction of the correct matches during deployment. To verify effectiveness of our results, we utilized two real world datasets in place recognition, both indoor and outdoor. Our investigation demonstrates that fine tuning architectures with ArcFace loss in an end-to-end manner outperforms other two losses by approximately 1~4% in outdoor and 1~2% in indoor datasets, given certain thresholds, for the visual place recognition tasks.