8 Papers
4 Citations
Min Li is an academic researcher from Beijing Jiaotong University. The author has contributed to research in topics: Object detection & Extreme point. The author has an hindex of 2, co-authored 8 publications. Previous affiliations of Min Li include Chinese Academy of Sciences.
Chat about Author
Papers
A Scale Balanced Loss for Bounding Box Regression
TL;DR: The scale balanced loss is proposed, which is asymmetric, position-sensitive, and scale-invariant, and designed as a fraction to eliminate the scale information contained in the numerator and denominator in order to solve the area maze problem.
Complete Video-Level Representations for Action Recognition
TL;DR: A method based on a 3D backbone network for multi scale spatial feature representation, which uses a pyramid pooling layer to allow the input of video frames at different scales, and then aggregates short-term spatial–temporal features into a long-term video-level representation is proposed.
Fusing RFID and Computer Vision for Occlusion-Aware Object Identifying and Tracking
Min Li,Yao Chen,Yanfang Zhang,Jian Yang,Hong Du +4 more
- 24 Jun 2019
TL;DR: Experimental results show that the proposed RFID and computer vision hybrid indoor tracking system can achieve 98% identification accuracy and centimeter-level tracking precision, even in long-term occlusion scenarios, which can manipulate various practical object-monitoring scenarios in the public security applications.
5
•Posted Content
GCsT: Graph Convolutional Skeleton Transformer for Action Recognition.
TL;DR: In this paper, the authors proposed a skeleton transformer based on graph convolutional networks (GCsT), which employs all the benefits of Transformer (i.e. dynamical attention and global context) while keeping the advantages of GCNs.
4
•Posted Content
Rethinking the Aligned and Misaligned Features in One-stage Object Detection.
TL;DR: Zhang et al. as mentioned in this paper proposed a simple and plug-in operator that can generate aligned and disentangled features for each task, respectively, without breaking the fully convolutional manner.
3