Zhenxiang Pan

Zhenxiang Pan

Auto-generated profile, based on publicly available information

Zhenxiang Pan

College of Computer Science and Software Engineering, Hohai University, Nanjing, China

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

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

Representative Publications

View All Publications
11 December 2025 in IEEE Transactions on Consumer Electronics
Verified
Journal Article
Closed access
TPTianfu Pang
Tianfu Pang
YMYingchi Mao
Yingchi Mao
ZPZhenxiang Pan
Zhenxiang Pan
RQRongzhi Qi
Rongzhi Qi
BZBenteng Zhang
Benteng Zhang
JWJie Wu
Jie Wu
Large Language Models (LLMs) can stimulate intelligent functions on mobile devices in the applications of consumer electronics. However, it is challenging to deploy LLMs on mobile devices due to limited memory and computational resources that fluctuate as devices execute multiple AI tasks concurrently. Quantization is a widely adopted compression technique, but faces two major difficulties in practical deployment: 1) Existing quantization methods are unable to select suitable models for deployment according to the resource limitations of mobile devices; 2) Mixed-precision quantization-based methods utilize reinforcement learning and element-level quantization, which incurs substantial computational and memory costs on mobile devices. To this end, we propose Resource Awareness-Based Salience-Weighted Mixed-Precision Quantization (RAS-MQ). Specifically, RAS-MQ constructs a memory and latency computation incorporating model parameters and device constraints, then RAS-MQ employs integer linear programming for optimal per-layer bit-width allocation. Weight matrices are partitioned into groups with bit-widths assigned based on average saliency ranking. Furthermore, RAS-MQ applies training-free quantization range clipping to salient weights, which can improve representational capacity while reducing quantization error. Experiments on five zero-shot inference datasets, for the 2-bit quantization of LLaMA2-13B, RAS-MQ outperforms the state-of-the-art baseline (SqueezeLLM) by 0.85% in accuracy and reduces weight memory consumption from 24.5 GB to 4.1 GB, achieving a 5.98× compression ratio.

1 January 2026 in IET Image Processing
Verified
Journal Article
Open access
PDF file
ZWZicheng Wang
Zicheng Wang
ZPZhenxiang Pan
Zhenxiang Pan
ZCZhiyu Cai
Zhiyu Cai
HZHui Zhou
Hui Zhou
YTYu Tang
Yu Tang
YMYingchi Mao
Yingchi Mao
Unmanned aerial vehicle (UAV) technology has achieved widespread applications. Object detection plays a crucial role in these applications. However, UAV‐based object detection faces three primary challenges: small object detection with complex backgrounds, novel category detection in open‐world scenarios and computational efficiency under resource constraints. Existing methods typically address these challenges individually, lacking unified solutions for resource‐constrained aerial platforms. We propose EAFE‐OSOD (efficient attention fusion encoder for open‐set object detection), an end‐to‐end framework that simultaneously tackles all three challenges through: (1) class‐agnostic mask embedding that guides the model to focus on specific regions, especially small objects; (2) CLIP text encoder integration for open‐set detection capabilities and (3) efficient fusion architectures that maintain computational efficiency through streamlined multi‐scale processing. Experiments on VisDroneZSD, VisDrone and UAVDT datasets demonstrate superior performance. On VisDrone, our method achieves 38.1% AP with 248.9G FLOPs, outperforming the second‐best method (36.6% AP and 658.7G FLOPs) while maintaining computational efficiency suitable for UAV platforms.

1 January 2026 in IEEE Transactions on Cognitive Communications and Networking
Verified
Journal Article
Closed access
BZBenteng Zhang
Benteng Zhang
YMYingchi Mao
Yingchi Mao
TPTianfu Pang
Tianfu Pang
ZPZhenxiang Pan
Zhenxiang Pan
JHJianxin Huang
Jianxin Huang
JWJie Wu
Jie Wu


3 August 2026
Verified
Book Chapter
Closed access
PDF file
JXJiawen Xue
Jiawen Xue
YTYu Tang
Yu Tang
YMYingchi Mao
Yingchi Mao
ZWZicheng Wang
Zicheng Wang
ZPZhenxiang Pan
Zhenxiang Pan
RQRongzhi Qi
Rongzhi Qi