Attentive Multimodal Fusion for Optical and Scene Flow

Proceedings Paper Computer Science, Artificial Intelligence

Multimodal Object Detection via Probabilistic Ensembling

Yi-Ting Chen et al.

Summary: This paper investigates multimodal object detection with RGB and thermal cameras and proposes a probabilistic ensembling technique, ProbEn, that effectively fuses the detections from different modalities. It shows significant improvement in multimodal detection even when the conditional independence assumption is not met.

COMPUTER VISION, ECCV 2022, PT IX (2022)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

CRAFT: Cross-Attentional Flow Transformer for Robust Optical Flow

Xiuchao Sui et al.

Summary: Optical flow estimation aims to find the 2D motion field by identifying corresponding pixels. This study proposes a new architecture called CRAFT, which improves the computation of correlations by transforming features and using Transformer cross-frame attention, leading to more accurate and robust optical flow estimation.

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022) (2022)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

CamLiFlow: Bidirectional Camera-LiDAR Fusion for Joint Optical Flow and Scene Flow Estimation

Haisong Liu et al.

Summary: This paper studies the problem of estimating optical flow and scene flow from synchronized 2D and 3D data. A novel end-to-end framework, called CamLiFlow, is proposed to address this problem. It consists of 2D and 3D branches with bidirectional connections and applies a point-based 3D branch to extract geometric features. Experimental results show that CamLiFlow achieves better performance with fewer parameters and ranks 1st on the KITTI Scene Flow benchmark.

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022) (2022)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

Separable Flow: Learning Motion Cost Volumes for Optical Flow Estimation

Feihu Zhang et al.

Summary: Full-motion cost volumes are crucial in optical flow methods, but their lack of encapsulating prior knowledge leads to artifacts in ambiguous regions. Our proposed separable cost volume module effectively utilizes global context clues and prior knowledge to accurately identify motions in these areas.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

Learning to Estimate Hidden Motions with Global Motion Aggregation

Shihao Jiang et al.

Summary: The paper discusses a method for solving occlusion problems more effectively in the two-frame scenario by modeling image self-similarities and introducing a global motion aggregation module to improve optical flow estimates. This approach achieved state-of-the-art results on the challenging Sintel dataset, significantly improving performance in occluded regions.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Enhancing optical-flow-based control by learning visual appearance cues for flying robots

G. C. H. E. de Croon et al.

Summary: Researchers have proposed a new method for robots to estimate distances between objects by their visual appearance, which has been successfully implemented on a small flying robot. This approach results in improved performance in tasks such as landing and obstacle avoidance.

NATURE MACHINE INTELLIGENCE (2021)

Add to Collection

Proceedings Paper Automation & Control Systems