01
Challenge Title
“AI+”-Enabled Fine-Grained Perception and Dynamic Assessment of Sudden-Onset Disasters Driven by Multimodal Ubiquitous Data
Challenge Description：This challenge explores innovative applications of multimodal large models in sudden-onset disaster response. Participants shall use multimodal unstructured data, including text, images, and videos, to develop an intelligent assessment system capable of extracting and quantifying key disaster indicators, thereby enabling timely situational awareness of disaster conditions.
Core Requirements：
1. Scenario Innovation: Develop innovative application scenarios for the early identification of disaster conditions, estimation of the affected area during extreme events, assessment of infrastructure damage, precise geolocation of public requests for assistance, or related needs.
2. Technology Integration: Achieve deep integration of multimodal large-model processing technologies and demonstrate the capability to handle complex backgrounds at disaster sites and non-standardized forms of expression.
3. Solution: Develop a complete technical solution.
4. Value Demonstration: Analyze the potential benefits of the proposed solution.
Submission Requirements：
1. Solution document: submit a PDF document of approximately 5,000 Chinese characters, including the scenario description and problem analysis, technical solution, implementation pathway, and value demonstration.
2. Project presentation: submit a PowerPoint file of no more than 40 slides.
3. Optional demonstration video: demonstrate the application outcomes of the proposed solution. The video shall not exceed eight minutes.
4. Optional prototype demonstration system: provide an accessible URL or a demonstration video.
5. Optional business plan: applicable to projects with potential for industrialization or commercialization.
02
Challenge Title
Integrated GIS Object Detection and Spatial Analysis Based on UAV Orthophotos
Challenge Description：For application scenarios integrating UAV aerial surveying, GIS, and deep learning, participants shall establish a technical pipeline for intelligent ground-object recognition, GIS spatial analysis, and application of results using high-resolution digital orthophoto map (DOM) data. Teams shall use UAV imagery, position and orientation system (POS) data, and ground-control-point data—either from public datasets or self-collected sources—to refine the DOM. Deep-learning techniques shall then be used for precise ground-object recognition, while GIS spatial-analysis algorithms shall support quantitative indicator calculation and thematic mapping, leading to a complete deployable application solution.
Core Requirements：
1. Overall Solution Design: The choice of software and hardware platforms is unrestricted. Based on the selected operational scenario, participants shall complete one of the following two solution-design options.
2. Data Acquisition and Preprocessing Solution: Specify the data source, whether a public dataset or self-collected data, and complete DOM production, image preprocessing, and standardization. When self-collected data are used, the proposed flight-route planning, ground-control-point deployment, and field aerial-survey implementation may also be briefly described.
3. Object Recognition, Change Detection, and Spatial Analysis Solution: Describe the selection of the deep-learning foundation model or pretrained model, the sample-acquisition and annotation strategy, model fine-tuning and inference methods, as well as indicator calculation and result-presentation methods for GIS spatial analysis.
Submission Requirements：
1. Solution document: submit a PDF document of approximately 5,000 Chinese characters, providing a complete account of the technical route, core parameters, accuracy-assessment methods, model design, GIS algorithm logic, and thematic-map design approach.
2. Project presentation: submit a PowerPoint file of no more than 40 slides, systematically presenting the problem analysis, technical solution, implementation process, key achievements, and application value.
3. Demonstration video: submit a video of no more than eight minutes that fully presents the entire workflow, including field aerial surveying, office-based data processing, model training, intelligent detection, GIS spatial analysis, and final output generation.
4. Optional prototype demonstration system: provide an accessible URL or a demonstration video.
5. Optional business plan: applicable to projects with potential for industrialization or commercialization.