Xiaokang Qu

Xiaokang Qu

Auto-generated profile, based on publicly available information

Xiaokang Qu

School of Cyber Science and Technology, University of Science and Technology of China, Hefei, Anhui, China

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

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

Representative Publications

View All Publications
12 January 2026 in IEEE Geoscience and Remote Sensing Magazine
Verified
Journal Article
Closed access
CWCongcong Wen
Congcong Wen
YLYiting Lin
Yiting Lin
XQXiaokang Qu
Xiaokang Qu
NLNan Li
Nan Li
YLYong LIAO
Yong LIAO
HLHui Lin
Hui Lin
Recent progress in vision-language models (VLMs) has demonstrated impressive capabilities across a variety of tasks in the natural image domain. Motivated by these advancements, the remote sensing community has begun to adopt VLMs for remote sensing vision-language tasks, including image captioning, scene understanding, and visual question answering (VQA). However, existing remote sensing VLMs typically rely on closed-set scene understanding and focus on generic scene descriptions, yet lack the ability to incorporate external knowledge. This limitation hinders their capacity for semantic reasoning over complex or context-dependent queries that involve domain-specific or world knowledge. To address these challenges, we first introduced a multimodal remote sensing world knowledge (RSWK) dataset, which comprises high-resolution satellite imagery and detailed textual descriptions for 14,820 well-known landmarks across 16 categories from 184 countries, integrating both remote sensing domain knowledge and broader world knowledge. Building upon this dataset, we proposed a novel remote sensing retrieval-augmented generation (RS-RAG) framework, which consists of two key components. The multimodal knowledge vector database (MKVD) construction module encodes remote sensing imagery and associated textual knowledge into a unified vector space. The knowledge retrieval and response generation module retrieves and reranks relevant knowledge based on image and/or text queries and incorporates the retrieved content into a knowledge-augmented prompt to guide the VLM in producing contextually grounded responses. We validate the effectiveness of RS-RAG on our designed benchmark covering three vision-language tasks: image captioning, image classification, and VQA, where RS-RAG significantly outperformed state-of-the-art baselines. By bridging remote sensing imagery and comprehensive knowledge, RS-RAG empowers remote sensing VLMs with enhanced contextual reasoning, enabling them to generate more accurate, informative, and semantically grounded outputs across a wide range of tasks.

28 August 2026 in bioRxiv
Verified
Preprint
Open access
PDF file
XQXiaokang Qu
Xiaokang Qu
CGChang Guo
Chang Guo
TFTianlong Fan
Tianlong Fan
Linyuan LüLinyuan Lü
Linyuan Lü
Species loss can erode food-web functioning not only through secondary extinctions, but also through biomass redistribution, weakened energy pathways, and threshold-like functional collapse. Common topology-, connectivity-, and extinction-based robustness metrics provide valuable summaries of structural disassembly and cascade risk, but they are not designed to quantify continuous biomass retention, collapse-associated species sets, and non-additive group-level effects within a single dynamic framework. We develop a dynamic biomass-based framework for assessing food-web robustness under progressive species removal. The framework introduces Dynamic Area-based Robustness (DAR), which quantifies the weighted area between slow- and fast-collapse reference trajectories of total ecosystem biomass retention. Building on these trajectories, we operationally define the Minimal Vital Species Set (MVSS) as the smallest fast-collapse-prefix species set whose removal first drives biomass below a predefined functional-collapse threshold. We further propose Cluster Influence (CI), which compares the biomass effect of simultaneous group removal with the mean effect of removing the same species individually. We evaluated the framework using 120 niche-model virtual food webs spanning controlled gradients of species richness and connectance, and further demonstrated its applicability on 16 empirical stream food webs. We compared DAR with AUC- and secondary-extinction-based robustness metrics and assessed the sensitivity of DAR, MVSS, and CI to key bioenergetic parameters and parameter uncertainty. DAR captured biomass-based robustness patterns that were only partly aligned with structural and extinction-based metrics, indicating that dynamic functional degradation provides complementary information. In virtual food webs, MVSS subsets were strongly enriched in basal species or basal resource nodes, and smaller MVSS proportions were associated with stronger positive CI under fast-collapse trajectories. Together, DAR, MVSS, and CI provide a reproducible framework for linking food-web structure, biomass dynamics, collapse thresholds, and non-additive species-set effects, offering a practical tool for dynamic robustness assessment in theoretical and empirical food webs.