CAMAL: Context-Aware Multi-layer Attention framework for Lightweight Environment Invariant Visual Place Recognition
In the last few years, Deep Convolutional Neural Networks (D-CNNs) have shown state-of-the-art (SOTA) performance for Visual Place Recognition (VPR), a pivotal component of long-term intelligent robotic vision (vision-aware localization and navigation systems). The prestigious generalization power of D-CNNs gained upon training on large scale places datasets and learned persistent image regions which are found to be robust for specific place recognition under changing conditions and camera viewpoints. However, against the computation and power intensive D-CNNs based VPR algorithms that are employed to determine the approximate location of resource-constrained mobile robots, lightweight VPR techniques are preferred. This paper presents a computation- and energy-efficient CAMAL framework that captures place-specific multi-layer convolutional attentions efficient for environment invariant-VPR. At 4x lesser power consumption, evaluating the proposed VPR framework on challenging benchmark place recognition datasets reveal better and comparable Area under Precision-Recall (AUC-PR) curves with approximately 4x improved image retrieval performance over the contemporary VPR methodologies.
Code (0)
등록된 구현이 없습니다.
Tasks
Image RetrievalRetrievalVisual Place RecognitionSimilar Papers 제목 키워드 기반
CAMAL: Improving Attention Alignment and Faithfulness with Segmentation Masks
Many vision datasets now provide segmentation masks in addition to annotated images to support a wide range of tasks. In this work, we propose Class Activation Map Attention Learning (CAMAL), an efficient and scalable me…
Reinforcement LearningCAMAL: Optimizing LSM-trees via Active Learning
We use machine learning to optimize LSM-tree structure, aiming to reduce the cost of processing various read/write operations. We introduce a new approach Camal, which boasts the following features: (1) ML-Aided: Camal i…
Active LearningFew Labels are all you need: A Weakly Supervised Framework for Appliance Localization in Smart-Meter Series
Improving smart grid system management is crucial in the fight against climate change, and enabling consumers to play an active role in this effort is a significant challenge for electricity suppliers. In this regard, mi…
AllNon-Intrusive Load MonitoringFlux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference
The quadratic computational complexity of standard attention mechanisms presents a severe scalability bottleneck for LLMs in long-context scenarios. While hybrid attention mechanisms combining Full Attention (FA) and Spa…
Mathematical ReasoningInformation RetrievalHierarchical Classification of Financial Transactions Through Context-Fusion of Transformer-based Embeddings and Taxonomy-aware Attention Layer
This work proposes the Two-headed DragoNet, a Transformer-based model for hierarchical multi-label classification of financial transactions. Our model is based on a stack of Transformers encoder layers that generate cont…
Hierarchical Multi-label ClassificationMulti-Label ClassificationMUlTI-LABEL-ClASSIFICATION