Author name cluster

Kefu Yi

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

2 papers

1 author row

EAAI Journal 2026 Journal Article

Multivariate feature learning and associative spatial information enhancement for snow object detection in autonomous driving

Jinlai Zhang
Mingchao Xiang
Yongheng Hu
Wei Hao
Linlong Lei
Kefu Yi

Object detection in autonomous driving systems based on artificial intelligence is particularly challenging in adverse weather conditions, with snow being one of the most severe. The occlusion and distortion caused by snowfall significantly degrade detection performance, leading to potential failures in critical decision-making processes. To address these challenges, Multivariate feature learning with Associative spatial information enhancement and Lighter fusion Detection model (MAL-Det) was introduced, a lightweight and efficient object detection model specifically designed for snow-affected environments and resource-constrained devices based on You Only Look Once version 8 small (YOLOv8s). MAL-Det incorporates three key innovations: Multivariate Feature Learning (MFL) Module, Associative Spatial Information (ASI) Enhancement Module and Lighter Fusion Cross Stage Partial 2 with Focus (LF_C2f) mechanism. Through extensive ablation studies on the real-world snowy object detection (RSOD) dataset, the efficacy of each module has been demonstrated. Comparative experiments with state-of-the-art (SOTA) models further validate MAL-Det’s superior performance in snow object detection, where MAL-Det achieves a mean Average Precision (mAP) of 59. 8% at Intersection over Union (IoU) 0. 50 and 40. 7% at IoU 0. 50-0. 95, positioning it as a promising solution for enhancing the safety and reliability of autonomous driving systems in snowy conditions.

Details DOI

AAAI Conference 2024 Conference Paper

UCMCTrack: Multi-Object Tracking with Uniform Camera Motion Compensation

Kefu Yi
Kai Luo
Xiaolei Luo
Jiangui Huang
Hao Wu
Rongdong Hu
Wei Hao

Multi-object tracking (MOT) in video sequences remains a challenging task, especially in scenarios with significant camera movements. This is because targets can drift considerably on the image plane, leading to erroneous tracking outcomes. Addressing such challenges typically requires supplementary appearance cues or Camera Motion Compensation (CMC). While these strategies are effective, they also introduce a considerable computational burden, posing challenges for real-time MOT. In response to this, we introduce UCMCTrack, a novel motion model-based tracker robust to camera movements. Unlike conventional CMC that computes compensation parameters frame-by-frame, UCMCTrack consistently applies the same compensation parameters throughout a video sequence. It employs a Kalman filter on the ground plane and introduces the Mapped Mahalanobis Distance (MMD) as an alternative to the traditional Intersection over Union (IoU) distance measure. By leveraging projected probability distributions on the ground plane, our approach efficiently captures motion patterns and adeptly manages uncertainties introduced by homography projections. Remarkably, UCMCTrack, relying solely on motion cues, achieves state-of-the-art performance across a variety of challenging datasets, including MOT17, MOT20, DanceTrack and KITTI. More details and code are available at https://github.com/corfyi/UCMCTrack.

PDF Details DOI