Abstract:Local features detection and description are widely used in many vision applications with high industrial and commercial demands. With large-scale applications, these tasks raise high expectations for both the accuracy and speed of local features. Most existing studies on local features learning focus on the local descriptions of individual keypoints, which neglect their relationships established from global spatial awareness. In this paper, we present AWDesc with a consistent attention mechanism (CoAM) that opens up the possibility for local descriptors to embrace image-level spatial awareness in both the training and matching stages. For local features detection, we adopt local features detection with feature pyramid to obtain more stable and accurate keypoints localization. For local features description, we provide two versions of AWDesc to cope with different accuracy and speed requirements. On the one hand, we introduce Context Augmentation to address the inherent locality of convolutional neural networks by injecting non-local context information, so that local descriptors can "look wider to describe better". Specifically, well-designed Adaptive Global Context Augmented Module (AGCA) and Diverse Surrounding Context Augmented Module (DSCA) are proposed to construct robust local descriptors with context information from global to surrounding. On the other hand, we design an extremely lightweight backbone network coupled with the proposed special knowledge distillation strategy to achieve the best trade-off in accuracy and speed. What is more, we perform thorough experiments on image matching, homography estimation, visual localization, and 3D reconstruction tasks, and the results demonstrate that our method surpasses the current state-of-the-art local descriptors. Code is available at: https://github.com/vignywang/AWDesc.

A novel biologically inspired local feature descriptor

A novel feature descriptor based on biologically inspired feature for head pose estimation.

Local Image Descriptor Inspired by Visual Cortex

A Novel Biologically-inspired Structural Model for Feature Correspondence

A Novel Local Invariant Feature Detection and Description Algorithm

LAP: a Bio-Inspired Local Image Structure Descriptor and Its Applications

Exploring Local Features and the Bag-of-Visual-Words Approach for Bioimage Classification.

Improved local descriptor (ILD): a novel fusion method in face recognition

Bio-inspired deep neural local acuity and focus learning for visual image recognition

A Bio-Inspired Image Local Feature Representation Based on Vision Network Model

A Novel Feature Descriptor Based Local Information for Image Retrieval

Local Salient Patterns &#x2014; A Novel Local Descriptor for Face Recognition

MTLDesc: Looking Wider to Describe Better

Attention Weighted Local Descriptors

A Novel Local Derivative Quantized Binary Pattern for Object Recognition.

Derf: Distinctive Efficient Robust Features From The Biological Modeling Of The P Ganglion Cells

Bundled Local Features for Image Representation

A review of local invariant features

Overlapping Local Image Descriptor

A novel facial appearance descriptor based on local binary pattern

A lightweight contour detection network inspired by biology