Dynamic Attention-based Visual Odometry

This paper proposes a dynamic attention-based visual odometry framework (DAVO), a learning-based VO method, for estimating the ego-motion of a monocular camera. DAVO dynamically adjusts the attention weights on different semantic categories for different motion scenarios based on optical flow maps. These weighted semantic categories can then be used to generate attention maps that highlight the relative importance of different semantic regions in input frames for pose estimation. In order to examine the proposed DAVO, we perform a number of experiments on the KITTI Visual Odometry and SLAM benchmark suite to quantitatively and qualitatively inspect the impacts of the dynamically adjusted weights on the accuracy of the evaluated trajectories. Moreover, we design a set of ablation analyses to justify each of our design choices, and validate the effectiveness as well as the advantages of DAVO. Our experiments on the KITTI dataset shows that the proposed DAVO framework does provide satisfactory performance in ego-motion estimation, and is able deliver competitive performance when compared to the contemporary VO methods.

Authors

• Xin-Yu Kuo
• Chien Liu
• Kai-Chen Lin
• Evan Luo
• Yu-Wen Chen
• Chun-Yi Lee

Keywords

• Simultaneous localization and mapping
• Semantics
• Pose estimation
• Benchmark testing
• Cameras
• Trajectory
• Visual odometry
• Important Regions
• Design Choices
• Optical Flow
• Attention Map
• Attention Weights
• Flow Map
• Input Frames
• Ablation Analysis
• KITTI Dataset
• Visual Framework
• Visual Simultaneous Localization And Mapping
• Semantic Regions
• Convolutional Layers
• Attention Mechanism
• Deep Convolutional Neural Network
• Semantic Segmentation
• Fully-connected Layer
• Attention Module
• Dynamic Weight
• Static Weight
• Consecutive Frames
• Translational Motion
• Depth Map
• Reference Ones
• Pixel Displacement
• Straight Road
• Accuracy Of Pose Estimation
• Global Pooling Layer