The team-led project “Development and Application of a Foundational Framework for Trustworthy 4D World Models” was approved under the Sichuan Major Science and Technology Special Program.
Aerospace Intelligence · Agentic AI · Embodied AI
MIG Research Group
Multimodal Intelligence Group (MIG) is affiliated with the Big Media Intelligence Team at the School of Computer Science and Engineering, University of Electronic Science and Technology of China (UESTC). Focusing on frontier research challenges in artificial intelligence and computer vision, MIG is dedicated to advancing fundamental theories and key technologies in intelligent visual perception, multimodal information fusion and understanding, and embodied intelligence.
We warmly welcome motivated students and early-career researchers who are passionate about artificial intelligence, computer vision, multimodal intelligence, and embodied intelligence to join our group, explore the frontiers of intelligent systems, and pursue innovative research together!
STAR-Planner Simulation
News
The team’s participating project “Key Technologies and Applications for Intelligent Air–Space–Ground–Network Collaborative Emergency Rescue” passed the CAAI scientific and technological achievement evaluation.
The team received an Excellence Award at the 6th Intelligent Unmanned Systems Application Challenge.
Jun Zhang’s paper “RCMNet: A relation-centric memory backbone for efficient remote sensing image understanding” was accepted by Knowledge-Based Systems.
Keli Wang’s paper “Label-Free Text Prototype Adaptation for Open Vocabulary Segmentation” was accepted to ECCV 2026.
Chaofan Qiao’s paper “Toward Real-World RAW Blind Super-Resolution: A Novel Metadata-Integrated Approach and Benchmark” was accepted by IEEE Transactions on Multimedia.
Yupeng Gao’s paper “Uncertainty-Guided Semi-Supervised Change Detection for Remote Sensing Images” was accepted by IEEE Transactions on Geoscience and Remote Sensing.
Mingfeng Zha’s paper “Hierarchical Consistency Learning for Test-time Adaptation in Camouflage Perception” was accepted by IEEE Transactions on Image Processing.
Yuhui Wu’s thesis “Research on Semantic-Guided Low-Light Image Enhancement Algorithms” was selected for the CSIG 2025 Master’s Thesis Incentive Program.
Yeran Zhang’s paper “DSPR: Dual-Stream Physics-Residual Networks for Trustworthy Industrial Time Series Forecasting” was accepted to ACM SIGKDD 2026.
Resource Overview
UAIV Low-Altitude Multimodal Dataset
A real-world low-altitude UAV dataset for urban intelligence, covering multi-scenario RGB/IR/metadata sensing, weather and illumination variation, change reasoning, Re-ID, restoration, and governance-oriented scene tasks.
UAIV Semi-Automated Annotation Platform
A low-altitude remote sensing annotation platform that connects metadata indexing, model pre-labeling, human review, OCR/VLM checks, event QA, and benchmark-ready export for UAIV-style dataset production.
Partners
中国航天科技集团
中国电科集团
中国航空工业集团
西湖大学
