arXiv:2511.19557v1 Announce Type: cross 
Abstract: Timely and accurate assessment of damages following natural disasters is essential for effective emergency response and recovery. Recent AI-based frameworks have been developed to analyze large volumes of aerial imagery collected by Unmanned Aerial Vehicles, providing actionable insights rapidly. However, creating and annotating data for training these models is costly and time-consuming, resulting in datasets that are limited in size and diversity. Furthermore, most existing approaches rely on traditional classification-based frameworks with fixed answer spaces, restricting their ability to provide new information without additional data collection or model retraining. Using pre-trained generative models built on in-context learning (ICL) allows for flexible and open-ended answer spaces. However, these models often generate hallucinated outputs or produce generic responses that lack domain-specific relevance. To address these limitations, we propose ThiFAN-VQA, a two-stage reasoning-based framework for visual question answering (VQA) in disaster scenarios. ThiFAN-VQA first generates structured reasoning traces using chain-of-thought (CoT) prompting and ICL to enable interpretable reasoning under limited supervision. A subsequent answer selection module evaluates the generated responses and assigns the most coherent and contextually accurate answer, effectively improve the model performance. By integrating a custom information retrieval system, domain-specific prompting, and reasoning-guided answer selection, ThiFAN-VQA bridges the gap between zero-shot and supervised methods, combining flexibility with consistency. Experiments on FloodNet and RescueNet-VQA, UAV-based datasets from flood- and hurricane-affected regions, demonstrate that ThiFAN-VQA achieves superior accuracy, interpretability, and adaptability for real-world post-disaster damage assessment tasks.

يمثل تقديم إطار العمل Think First, Assign Next (ThiFAN-VQA) تقدمًا كبيرًا في تقييم الأضرار بعد الكوارث، حيث يستخدم الذكاء الاصطناعي لتحليل الصور الجوية الملتقطة بواسطة الطائرات بدون طيار (UAV) للحصول على رؤى دقيقة وفي الوقت المناسب. تهدف هذه الطريقة ذات المرحلتين إلى التغلب على القيود الموجودة في النماذج الحالية التي تعتمد على مساحات إجابات ثابتة وتتطلب جمع بيانات مكثف للتدريب.

La introducción del marco Think First, Assign Next (ThiFAN-VQA) marca un avance significativo en la evaluación de daños tras desastres, utilizando IA para analizar imágenes aéreas de Vehículos Aéreos No Tripulados (UAV) para obtener información precisa y oportuna. Este enfoque de dos etapas busca superar las limitaciones de los modelos existentes que dependen de espacios de respuesta fijos y requieren una extensa recolección de datos para su entrenamiento.

L'introduction du cadre Think First, Assign Next (ThiFAN-VQA) représente une avancée significative dans l'évaluation des dommages après une catastrophe, utilisant l'IA pour analyser les images aériennes provenant de véhicules aériens sans pilote (UAV) afin d'obtenir des informations précises et rapides. Cette approche en deux étapes vise à surmonter les limitations des modèles existants qui reposent sur des espaces de réponses fixes et nécessitent une collecte de données extensive pour l'entraînement.

The introduction of the Think First, Assign Next (ThiFAN-VQA) framework marks a significant advancement in post-disaster damage assessment, utilizing AI to analyze aerial imagery from Unmanned Aerial Vehicles (UAVs) for timely and accurate insights. This two-stage chain-of-thought approach aims to overcome limitations in existing models that rely on fixed answer spaces and require extensive data collection for training.

Think First, Assign Next (ThiFAN-VQA): A Two-stage Chain-of-Thought Framework for Post-Disaster Damage Assessment

arXiv:2512.07385v1 Announce Type: new 
Abstract: Unmanned Aerial Vehicles (UAVs) offer wide-ranging applications but also pose significant safety and privacy violation risks in areas like airport and infrastructure inspection, spurring the rapid development of Anti-UAV technologies in recent years. However, current Anti-UAV research primarily focuses on RGB, infrared (IR), or RGB-IR videos captured by fixed ground cameras, with little attention to tracking target UAVs from another moving UAV platform. To fill this gap, we propose a new multi-modal visual tracking task termed UAV-Anti-UAV, which involves a pursuer UAV tracking a target adversarial UAV in the video stream. Compared to existing Anti-UAV tasks, UAV-Anti-UAV is more challenging due to severe dual-dynamic disturbances caused by the rapid motion of both the capturing platform and the target. To advance research in this domain, we construct a million-scale dataset consisting of 1,810 videos, each manually annotated with bounding boxes, a language prompt, and 15 tracking attributes. Furthermore, we propose MambaSTS, a Mamba-based baseline method for UAV-Anti-UAV tracking, which enables integrated spatial-temporal-semantic learning. Specifically, we employ Mamba and Transformer models to learn global semantic and spatial features, respectively, and leverage the state space model's strength in long-sequence modeling to establish video-level long-term context via a temporal token propagation mechanism. We conduct experiments on the UAV-Anti-UAV dataset to validate the effectiveness of our method. A thorough experimental evaluation of 50 modern deep tracking algorithms demonstrates that there is still significant room for improvement in the UAV-Anti-UAV domain. The dataset and codes will be available at {\color{magenta}https://github.com/983632847/Awesome-Multimodal-Object-Tracking}.

تقدم دراسة جديدة مهمة تتعلق بمهمة تتبع مرئي متعددة الوسائط تُعرف باسم UAV-Anti-UAV، والتي تركز على تحدي تتبع طائرة مسيرة مستهدفة من منصة طائرة مسيرة أخرى. تعالج هذه المهمة فجوة كبيرة في الأبحاث الحالية المتعلقة بمكافحة الطائرات المسيرة، والتي اعتمدت بشكل أساسي على الكاميرات الثابتة على الأرض ووسائط الفيديو التقليدية. تقدم الدراسة مجموعة بيانات تتكون من مليون فيديو لدعم هذا المجال البحثي.

Un nuevo estudio presenta una tarea de seguimiento visual multimodal llamada UAV-Anti-UAV, centrada en el desafío de rastrear un UAV objetivo desde otra plataforma UAV. Esta tarea aborda una brecha significativa en la investigación actual sobre Anti-UAV, que se ha centrado principalmente en cámaras fijas en tierra y modalidades de video tradicionales. El estudio presenta un conjunto de datos a escala de un millón de videos para apoyar esta área de investigación.

Une nouvelle étude présente une tâche de suivi visuel multimodal appelée UAV-Anti-UAV, axée sur le défi de suivre un UAV cible depuis une autre plateforme UAV. Cette tâche comble une lacune significative dans la recherche actuelle sur l'Anti-UAV, qui s'est principalement appuyée sur des caméras fixes au sol et des modalités vidéo traditionnelles. L'étude propose un ensemble de données à l'échelle d'un million de vidéos pour soutenir ce domaine de recherche.

A new study introduces a multi-modal visual tracking task called UAV-Anti-UAV, focusing on the challenge of tracking a target UAV from another UAV platform. This task addresses a significant gap in current Anti-UAV research, which has primarily relied on fixed ground cameras and traditional video modalities. The study presents a million-scale dataset of 1,810 videos to support this research area.

Think First, Assign Next (ThiFAN-VQA): A Two-stage Chain-of-Thought Framework for Post-Disaster Damage Assessment

Was this article worth reading? Share it

LucidQuery AI

Attentive AI

Dyad