Jiangxi University of Finance and Economics
JUFE-EVL Lab
Computer vision & machine learning research group
We study visual perception and structured representation — from object pose and deformable object categories to large-scale dataset construction. This page is the home of our lab on GitHub; our datasets, toolkits and paper pages live here.
What we work on
Object pose estimation
Keypoint-based pose estimation for extensive object classes, built on the structure-prototype view of deformable categories.
Large-scale datasets
Dataset construction and annotation methodology — see our latest release, PoseImageNet.
Structured representations
Prototypes, skeletons and deformable models as intermediate structure for recognition and generation.
Selected publications
2026
-
PoseImageNet: Pose Estimation for Extensive Classes Based on Rich Structure Prototypes
-
Learning and Aligning Click-Aware Shape Prior for Interactive Amodal Instance Segmentation
-
Learning to Propose Pose for Category-Agnostic Objects via Joint Refinement with Co-Matching Supervision
-
Point-supervised Oriented Object Detection Baseline with Placement Prior
2025
-
Recurrent Feature Mining and Keypoint Mixup Padding for Category-Agnostic Pose Estimation
-
Webly Supervised Fine-Grained Classification by Integrally Tackling Noises and Subtle Differences
2024
-
Meta-Point Learning and Refining for Category-Agnostic Pose Estimation
Datasets
-
Completed-UniKPT: A Completed Multi-Class Pose Dataset
Titles follow the official GitHub repository descriptions; preprint links will be added as they become available.
People
The lab is led by Junjie Chen (Jiangxi University of Finance and Economics) — see the full member list and group activities on the group page.
Get in touch
Email: [to be added] · Address: Jiangxi University of Finance and Economics, Nanchang, China