Transfer Learning for Wildfire Identification in UAV Imagery

Wu, Haiyu; Li, Huayu; Shamsoshoara, Alireza; Razi, Abolfazl; Afghah, Fatemeh

doi:10.1109/ciss48834.2020.1570617429

Cited by 34 publications

(14 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most supervised learning methods rely on large training datasets to train a reasonably accurate model. Studies such as [30] used a fire dataset from public sources to perform fire detection based on pre-trained ANN architectures such as MobileNet and AlexNet. However, that dataset was based on terrestrial images of the fire.…”

mentioning

confidence: 99%

Aerial Imagery Pile burn detection using Deep Learning: the FLAME dataset

Shamsoshoara¹,

Afghah²,

Razi³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Wildfires are one of the costliest and deadliest natural disasters in the US, causing damage to millions of hectares of forest resources and threatening the lives of people and animals. Of particular importance are risks to firefighters and operational forces, which highlights the need for leveraging technology to minimize danger to people and property. FLAME (Fire Luminosity Airbornebased Machine learning Evaluation) offers a dataset of aerial images of fires along with methods for fire detection and segmentation which can help firefighters and researchers to develop optimal fire management strategies. This paper provides a fire image dataset collected by drones during a prescribed burning piled detritus in an Arizona pine forest. The dataset includes video recordings and thermal heatmaps captured by infrared cameras. The captured videos and images are annotated, and labeled frame-wise to help researchers easily apply their fire detection and modeling algorithms. The paper also highlights solutions to two machine learning problems: (1) Binary classification of video frames based on the presence [and absence] of fire flames. An Artificial Neural Network (ANN) method is developed that achieved a 76% classification accuracy. (2) Fire detection using segmentation methods to precisely determine fire borders. A deep learning method is designed based on the U-Net

show abstract

mentioning

confidence: 99%

Aerial Imagery Pile burn detection using Deep Learning: the FLAME dataset

Shamsoshoara¹,

Afghah²,

Razi³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Using transfer learning and data augmentation strategies, this method achieved an accuracy of 99.3%. It outperformed published state-of-the-art methods such as Fire_Net and AlexNet and proved its suitability in detecting forest fire on aerial monitoring systems [38].…”

mentioning

confidence: 86%

“…Using the UAV_Fire dataset [36], Fire_Net achieved an accuracy of 98% and outperformed previous methods. Wu et al [38] used a pretrained MobileNetv2 [47] model to detect both smoke and fire. MobileNetv2 is an extended version of MobileNetv1 [48], which is a lightweight CNN with depth-wise separable convolutions.…”

mentioning

confidence: 99%

Deep Learning and Transformer Approaches for UAV-Based Wildfire Detection and Segmentation

Ghali

Akhloufi

Mseddi

2022

Sensors

View full text Add to dashboard Cite

Wildfires are a worldwide natural disaster causing important economic damages and loss of lives. Experts predict that wildfires will increase in the coming years mainly due to climate change. Early detection and prediction of fire spread can help reduce affected areas and improve firefighting. Numerous systems were developed to detect fire. Recently, Unmanned Aerial Vehicles were employed to tackle this problem due to their high flexibility, their low-cost, and their ability to cover wide areas during the day or night. However, they are still limited by challenging problems such as small fire size, background complexity, and image degradation. To deal with the aforementioned limitations, we adapted and optimized Deep Learning methods to detect wildfire at an early stage. A novel deep ensemble learning method, which combines EfficientNet-B5 and DenseNet-201 models, is proposed to identify and classify wildfire using aerial images. In addition, two vision transformers (TransUNet and TransFire) and a deep convolutional model (EfficientSeg) were employed to segment wildfire regions and determine the precise fire regions. The obtained results are promising and show the efficiency of using Deep Learning and vision transformers for wildfire classification and segmentation. The proposed model for wildfire classification obtained an accuracy of 85.12% and outperformed many state-of-the-art works. It proved its ability in classifying wildfire even small fire areas. The best semantic segmentation models achieved an F1-score of 99.9% for TransUNet architecture and 99.82% for TransFire architecture superior to recent published models. More specifically, we demonstrated the ability of these models to extract the finer details of wildfire using aerial images. They can further overcome current model limitations, such as background complexity and small wildfire areas.

show abstract

“…To enable smart farming, UAVs can collect field-level data such as plant count, soil H 2 O level, temperature or imagery data for plant and animal monitoring [157]- [160]. UAVs can also detect the growth of the plants [161], identify diseases in advance, reduce crop damage [162], support emergency situations that can harm the farms, such as wildfire [163], help with planting, crop spraying and irrigation [164], [165], fruit counting [166], [167] and planned harvesting [168]. All these use cases can advance agriculture in the years to come.…”

Section: Uavmentioning

confidence: 99%

A Survey of Wireless Networks for Future Aerial COMmunications (FACOM)

Baltaci,

Dinc,

Ozger

et al. 2021

Preprint

View full text Add to dashboard Cite

Electrification turned over a new leaf in aviation by introducing new types of aerial vehicles along with new means of transportation. Addressing a plethora of use cases, drones are gaining attention in the industry and increasingly appear in the sky. Emerging concepts of flying taxi enable passengers to be transported over several tens of kilometers. Therefore, unmanned traffic management systems are under development to cope with the complexity of future airspace, thereby resulting in unprecedented communication needs. Moreover, the long-term increase in the number of commercial airplanes pushes the limits of voice-oriented communications, and future options such as single-pilot operations demand robust connectivity. In this survey, we provide a comprehensive review and vision for enabling the connectivity applications of aerial vehicles utilizing current and future communication technologies. We begin by categorizing the connectivity use cases per aerial vehicle and analyzing their connectivity requirements. By reviewing more than 500 related studies, we aim for a comprehensive approach to cover wireless communication technologies, and provide an overview of recent findings from the literature toward the possibilities and challenges of employing the wireless communication standards. After analyzing the proposed network architectures, we list the open-source testbed platforms to facilitate future investigations by researchers. This study helped us observe that while numerous works focused on cellular technologies to enable connectivity for aerial platforms, a single wireless technology is not sufficient to meet the stringent connectivity demands of the aerial use cases, especially for the piloting operations. We identified the need of further investigations on multi-technology heterogeneous network architectures to enable robust and real-time connectivity in the sky. Future works should consider suitable technology combinations to develop unified aerial networks that can meet the diverse quality of service demands of the aerial use cases.

show abstract

Transfer Learning for Wildfire Identification in UAV Imagery

Cited by 34 publications

References 16 publications

Aerial Imagery Pile burn detection using Deep Learning: the FLAME dataset

Aerial Imagery Pile burn detection using Deep Learning: the FLAME dataset

Deep Learning and Transformer Approaches for UAV-Based Wildfire Detection and Segmentation

A Survey of Wireless Networks for Future Aerial COMmunications (FACOM)

Contact Info

Product

Resources

About