Dong Feng

Dong Feng

Auto-generated profile, based on publicly available information

Dong Feng

Inchitech

Research Focus & Metrics

Academic Career Length

29Years (1998-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 April 2021 in IEEE Access
Verified
Journal Article
Open access
DFDong Feng
Dong Feng
ZWZhongcheng Wu
Zhongcheng Wu
JZJun Zhang
Jun Zhang
TRTingting Ren
Tingting Ren
Graph convolutional networks (GCNs) have achieved remarkable performance on skeleton-based action recognition. Existing GCN-based methods usually apply the fixed graph topology and one fixed temporal convolution kernel to extract the spatial features of joints and temporal features, which is from a single-scale perspective. Actually, human actions are coordinated by various body parts in the spatial domain, and exhibit different characteristics in the temporal domain. Therefore, it is appropriate to model the multi-scale information that can enhance both the explainability and stability, which is ignored in current literatures. To address this issue, we propose a multi-scale spatial-temporal graph neural network (MSTGNN) to discover multi-scale discriminative features from spatial and temporal aspects simultaneously. Our contributions are three-folds: 1) For the spatial domain, inspired by the kinematics of the human action, we develop a three-scale graph data structures in a fine-to-coarse way. A novel hybrid spatial pooling module is then proposed to dynamically exploit the global and comprehensive information step-by-step. 2) For the temporal domain, we design a multi-scale temporal convolution module adaptively fusing the temporal features extracted by different scale convolution kernels. 3) As utilizing one-stream architecture instead of multi-stream architecture, the proposed model can be trained in an end-to-end manner. MSTGNN achieves state-of-the-art performance with less computation complexity. Experimental results conducted on two large datasets (NTU-RGB+D and NTU-RGB+D-120) demonstrate the superiority of MSTGNN.

17 November 2020 in IEEE Access
Verified
Journal Article
Open access
DFDong Feng
Dong Feng
ZWZhongcheng Wu
Zhongcheng Wu
JZJun Zhang
Jun Zhang
ZWZiheng Wu
Ziheng Wu
Predicting traffic speed accurately is a very challenging task of the intelligent traffic system (ITS), due to the complex and dynamic spatial-temporal dependencies from both temporal and spatial aspects. There not only exits short-term local neighboring fluctuation and long-term global trend in temporal aspect, but also local and global correlations in spatial aspect. Most existing work focus on the local spatial-temporal dependencies, ignoring the global dynamic spatial-temporal corrections, which is comparably critical for traffic speed prediction. To address this problem, we propose a novel Dynamic Global-Local Spatial-Temporal Network(DGLSTNet) for traffic speed prediction, which consists of multiple spatial-temporal module considering the local and global information simultaneously from both temporal and spatial perspective. Each temporal module applies stacked dilated convolution block to exploit multi-scale local temporal information. Moreover, we empoly a global temporal attention block to capture global dependencies of temporal domain in an attention mechanism. In each spatial module, we not only learn the local but also focus on dynamic global spatial information learned by dymamic graph learning block. Combining the feature results from local and global perspective, the capability and expressiveness of traffic predicting model is improved. Experiment results on two real-world traffic datasets have demonstrated that our proposed model can effectively capture the comprehensive spatial-temporal dependencies and can achieve state-of-the-art prediction performance compared with the existing works.

1 April 1998 in Journal of Fashion Marketing and Management: An International Journal
Verified
Journal Article
Closed access
PDF file
WAWinifred Aldrich
Winifred Aldrich
BSBrian Smith
Brian Smith
FDFeng Dong
Feng Dong
This paper describes research undertaken at the Nottingham Trent University which investigated body movements and their relationship to garment design. The study identifies the difference between ergonomic measuring positions and the natural postures used by real figures in real activities. A new approach to the identification and coding of upper body postures has been made. A body coding system and a simple piece of equipment was designed that enables extended natural body positions to be recorded, thus achieving repeatability. This work enabled comparisons of aesthetic appearance and the functional comfort of women's tailored jackets to be examined.

19 April 2009 in The Journal of Supercomputing
Verified
Journal Article
Closed access
PDF file
AAAnupam Agrawal
Anupam Agrawal
JKJosef Kohout
Josef Kohout
GCGordon J. Clapworthy
Gordon J. Clapworthy
NMNigel J. B. McFarlane
Nigel J. B. McFarlane
FDFeng Dong
Feng Dong
DTDebora Testi
Debora Testi

11 April 2025 in Proceedings of the AAAI Conference on Artificial Intelligence
Verified
Journal Article
Open access
PDF file
DFDong Feng
Dong Feng
PGPing Guo
Ping Guo
EPEncheng Peng
Encheng Peng
MZMingmin Zhu
Mingmin Zhu
WYWenhao Yu
Wenhao Yu
PWPeng Wang
Peng Wang
Manipulating human poses based on natural language is an emerging research field that has traditionally focused on coarse commands such as “walking” or “dancing.” However, fine-grained pose manipulation, like instructing “put both hands in front of the stomach,” remains underexplored. In this paper, we introduce PoseLLaVA, a pioneering model that integrates SMPL-based pose representations into the multimodal LLaVA framework. Through a novel pose encoder decoder mechanism, PoseLLaVA achieves precise alignment between pose, textual, and visual modalities, enabling detailed control over pose manipulation tasks. PoseLLaVA excels in three key tasks: pose estimation, generation, and adjustment, all driven by detailed language instructions. We further introduce a fine-grained pose adjustment dataset PosePart, where each sample contains an initial pose and a target pose, along with specific instructions for adjustments, mimicking the guidance a human instructor might provide. Extensive evaluations across these tasks demonstrate significant improvements over existing methods, including metrics such as MPJPE and PA-MPJPE, which measure SMPL reconstruction errors, and Recall rates, which assess feature alignment across modalities. Specifically, PoseLLaVA reduces MPJPE errors by more than 20% compared to state-of-the-art methods in pose adjustment and generation tasks. Additionally, we demonstrate the feasibility of combining PoseLLaVA with generative models, such as diffusion, for pose image editing, highlighting its potential applications in language-controlled pose manipulation.