SAMFA: A Flame Segmentation Algorithm for Infrared and Visible Aerial Images in the Same Scene

Drones, Journal Year: 2025, Volume and Issue: 9(3), P. 217 - 217

Published: March 18, 2025

Existing aerial forest fire monitoring data primarily consist of infrared or visible light images. However, there is a lack in-depth research on the ability models to perceive regions across different spectral To address this, we first constructed dataset and images captured in same scene, from perspective, at time, with pixel-level segmentation annotations flame In response issues poor performance current large number learnable parameters models, propose an improved model algorithm, SAMFA (Segmentation Anything Model, Fire, Adapter). Firstly, while freezing original model, only additionally incorporated Adapter module fine-tuned better adapt network specificities task. Secondly, enhance network’s perception edges, U-shaped mask decoder designed. Lastly, reduce training difficulty, progressive strategy combining self-supervised fully supervised learning employed optimize entire model. We compared five state-of-the-art image algorithms labeled public dataset, experimental results demonstrate that performs best. Compared SAM, improves IoU by 11.94% 6.45% images, respectively, reducing 11.58 M.

Language: Английский

---

Meta Heuristic Image Optimisation in LiDAR Sensor Based on Cloud IoT Network and Deep Learning Algorithms

Remote Sensing in Earth Systems Sciences, Journal Year: 2025, Volume and Issue: unknown

Published: March 25, 2025

Language: Английский

---

Impact of LiDAR pulse density on forest fuels metrics derived using LadderFuelsR

Ecological Informatics, Journal Year: 2025, Volume and Issue: unknown, P. 103135 - 103135

Published: April 1, 2025

Language: Английский

---