Learning Context Flexible Attention Model for Long-Term Visual Place Recognition

Chen, Zetao; Liu, Lingqiao; Sa, Inkyu; Ge, Zongyuan; Chli, Margarita

doi:10.1109/lra.2018.2859916

Cited by 105 publications

(99 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Other than environmental variations, texture-less and low-informative scenes also pose difficulty to place matching. We show examples of all these challenges taken from public VPR datasets [2], [3], [4] in Fig. 1.…”

Section: Introductionmentioning

confidence: 99%

CoHOG: A Light-Weight, Compute-Efficient, and Training-Free Visual Place Recognition Technique for Changing Environments

Zaffar

Ehsan

Milford

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

This letter presents a novel, compute-efficient and training-free approach based on Histogram-of-Oriented-Gradients (HOG) descriptor for achieving state-of-the-art performance-per-compute-unit in Visual Place Recognition (VPR). The inspiration for this approach (namely CoHOG) is based on the convolutional scanning and regions-based feature extraction employed by Convolutional Neural Networks (CNNs). By using image entropy to extract regions-of-interest (ROI) and regional-convolutional descriptor matching, our technique performs successful place recognition in changing environments. We use viewpoint-and appearance-variant public VPR datasets to report this matching performance, at lower RAM commitment, zero training requirements and 20 times lesser feature encoding time compared to state-of-the-art neural networks. We also discuss the image retrieval time of CoHOG and the effect of CoHOG's parametric variation on its place matching performance and encoding time. Index Terms-SLAM, visual place recognition, autonomous vehicle navigation, computer vision for automation.

show abstract

Section: Introductionmentioning

confidence: 99%

CoHOG: A Light-Weight, Compute-Efficient, and Training-Free Visual Place Recognition Technique for Changing Environments

Zaffar

Ehsan

Milford

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

show abstract

“…Chen et al in [7] presented a VPR approach that identifies pivotal landmarks by directly extracting prominent patterns based on responses of late convolutional layers of deep object-centric VGG-16 model. Recently, Chen et al in [8] introduced a contextflexible attention model and combined it with a pre-trained objectcentric VGG-16 fine-tuned on SPED [24] to learn more powerful condition-invariant regional features. The system has shown state-ofthe-art performance on severe condition-variant datasets.…”

Section: Literature Reviewmentioning

confidence: 99%

“…Using the learned codebook, F L regions of benchmark test / reference traverses are quantized in (7) to predict the clusters or labels Z L , where α is the quantization function. Employing regions-based F L feature, predicted labels Z L and regional codebook C L , summed residue v corresponding to each u th region can be retrieved using (8).…”

Section: Regional Vocabulary and Extraction Of Vlad For Image Matcmentioning

confidence: 99%

“…The Synthesized Nordland dataset was recorded on a train journey with one traverse taken in winter and the other traverse was recorded in spring. Viewpoint variance was added by cropping frames of summer traverse to keep 75% resemblance [8]. For Berlin A100, Berlin Halenseestrasse and Berlin Kudamm, geotagged information is used for ground truth with 0 to ±2 frame tolerance.…”

Section: A Benchmark Place Recognition Datasetsmentioning

confidence: 99%

See 1 more Smart Citation

A Holistic Visual Place Recognition Approach Using Lightweight CNNs for Significant ViewPoint and Appearance Changes

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Despite their success, these approaches typically rely on the usage of CNN models which are pretrained on various computer vision datasets [41] using millions of images [5], [6], [38]. Training CNN models in an end-to-end fashion specifically for VPR have also recently been proposed [4], [38], [42]. However, they are still using common network architectures, i.e., AlexNet [43], VGG [44], ResNet [45], with slight changes to perform VPR.…”

Section: B Deep Neural Network For Visual Place Recognitionmentioning

confidence: 99%

A Hybrid Compact Neural Architecture for Visual Place Recognition

Chancán

Hernandez-Nuñez

Narendra

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

State-of-the-art algorithms for visual place recognition, and related visual navigation systems, can be broadly split into two categories: computer-science-oriented models including deep learning or image retrieval based techniques with minimal biological plausibility, and neuroscience-oriented dynamical networks that model temporal properties found in neural cells underlying spatial navigation in the brain. In this paper, we propose a new compact and high-performing place recognition hybrid model that bridges this divide for the first time. Our approach comprises two key components that incorporate neural models of these two categories: (1) FlyNet, a compact, sparse twolayer neural network inspired by brain architectures of fruit flies, Drosophila melanogaster, and (2) a one-dimensional continuous attractor neural network (CANN). The resulting FlyNet+CANN network combines the compact pattern recognition capabilities of our FlyNet model with the powerful temporal filtering capabilities of an equally compact CANN, replicating entirely in a hybrid neural implementation the functionality that yields high performance in algorithmic localization approaches like SeqSLAM. We evaluate our approach, and compare it to three state-of-the-art place recognition methods, on two benchmark real-world datasets with small viewpoint variations and extreme environmental changes; including day/night cycles where it achieves an AUC performance of 87% compared to 60% for Multi-Process Fusion, 46% for LoST-X and 1% for SeqSLAM, while being 6.5, 310, and 1.5 times faster respectively.

show abstract

Learning Context Flexible Attention Model for Long-Term Visual Place Recognition

Cited by 105 publications

References 39 publications

CoHOG: A Light-Weight, Compute-Efficient, and Training-Free Visual Place Recognition Technique for Changing Environments

CoHOG: A Light-Weight, Compute-Efficient, and Training-Free Visual Place Recognition Technique for Changing Environments

A Holistic Visual Place Recognition Approach Using Lightweight CNNs for Significant ViewPoint and Appearance Changes

A Hybrid Compact Neural Architecture for Visual Place Recognition

Contact Info

Product

Resources

About