An open flame and smoke detection dataset for deep learning in remote sensing based fire detection

Ming Wang,Peng Yue,Liangcun Jiang,Dayu Yu,Tianyu Tuo,Jian Li

DOI: https://doi.org/10.1080/10095020.2024.2347922

IF: 4.278

2024-06-07

Geo-spatial Information Science

Abstract:The early warning of fires is pivotal for preventing substantial economic losses and ecological damage. However, the enhancement of fire model generalization performance still faces challenges due to limitations in existing fire datasets, such as image quantity and heterogeneity. Taking the advantage of rapid advancements in remote sensing technology and artificial intelligence, we introduce the Flame And Smoke Detection Dataset (FASDD), a pioneering collection comprising over 120,000 heterogeneous images covering various fire scenarios, aiming to promote the evolution of fire detection models. FASDD serves as a challenging benchmark for object detection tasks and features three sub-datasets: FASDD_CV, FASDD_UAV, and FASDD_RS. These sub-datasets comprise cross-domain images from ground-based, airborne, and spaceborne sensors, respectively. Extensive experiments using advanced Swin Transformer models demonstrate satisfactory fire detection performance, with mAP scores of 84.9%, 89.7%, and 74.0% on the respective sub-datasets. Notably, the application case in wildfire localization highlights the proficiency of FASDD_RS models for monitoring and tracking forest fires. Trained on FASDD_RS, FASDD_UAV, and FASDD_CV, deep learning models can be individually deployed on edge devices including satellites, drones, and terrestrial sensors, thus empowering collaborative fire warning within the space-air-ground integrated observation network. Additionally, a standardized procedure for data preparation, data procedure, and quality assurance is established, providing off-the-shelf annotations with four common configurations to facilitate machine learning model training. FASDD stands as a pivotal benchmark, providing valuable resources for training and evaluating deep learning fire detection models, especially Large Vision Models (LVM). It holds the potential to accelerate research progress in urban firefighting or forest fire detection, providing vital security measures for early fire warning and emergency response tasks.

remote sensing

What problem does this paper attempt to address?

The paper primarily aims to address the challenges in early fire warning, particularly by enhancing the generalization performance of deep learning-based fire detection models. To overcome the limitations of existing fire datasets in terms of image quantity, diversity, and complexity, the authors have constructed a new dataset named Flame And Smoke Detection Dataset (FASDD).

FASDD is a pioneering dataset containing over 120,000 heterogeneous images covering various fire scenarios. It aims to facilitate the development of fire detection models and serves as a challenging benchmark for object detection tasks. FASDD consists of three sub-datasets: FASDD_CV (images from ground sensors), FASDD_UAV (images from drones), and FASDD_RS (images from satellites). These sub-datasets respectively include images from different domains, namely ground-based, aerial-based, and space-based sensors.

The paper also mentions extensive experiments using the advanced Swin Transformer model to validate the effectiveness of the FASDD dataset. The experimental results show that the model achieved satisfactory fire detection performance on these three sub-datasets, with mean Average Precision (mAP) of 84.9%, 89.7%, and 74.0%, respectively.

Additionally, the paper highlights several key features of the FASDD dataset:
1. **Large-scale**: FASDD is currently the largest and most comprehensive publicly available fire detection object detection dataset, containing over 120,000 images and 180,000 meticulously annotated instances.
2. **Diverse sample characteristics**: FASDD exhibits extensive variations in terms of image size, resolution, lighting intensity, viewing distance, sensor type, scene, and data source.
3. **Significant intra-class diversity and inter-class similarity**: Flame or smoke objects show diversity in size, shape, and color, while there is also a certain similarity between flames and smoke, such as smoke appearing red under flame illumination.

In summary, the goal of this paper is to improve the performance and generalization ability of fire detection algorithms by constructing a large-scale, diverse fire detection dataset, thereby enhancing the accuracy of early fire warning.