What question did this study set out to answer?

To develop a lightweight, high-speed deep learning model based on YOLOv7 that accurately detects forest fire smoke from unmanned aerial vehicles while minimizing false alarms from cloud interference.

July 30, 2023Remote SensingOpen Access

LMDFS: A Lightweight Model for Detecting Forest Fire Smoke in UAV Images Based on YOLOv7

View Full Paper

Ask AI

Bookmark

Authors

CGChen GongHong Kong University of Science and Technology RCRenxi ChengSoutheast University XLXufeng LinNanjing Forestry University

Discussion

Member takes

Implication

Model evaluation study demonstrates improved smoke detection in aerial imagery, suggesting practical feasibility for early wildfire suppression via drones.

Key Points

To develop a lightweight, high-speed deep learning model based on YOLOv7 that accurately detects forest fire smoke from unmanned aerial vehicles while minimizing false alarms from cloud interference.
Built a specialized forest smoke dataset incorporating crawled web imagery and synthetic smoke generation techniques.
Integrated Ghost Shuffle Convolution (GSConv) modules (GSELAN and GSSPPFCSPC), Coordinate Attention (CA), CARAFE upsampling, and SIoU loss into the YOLOv7 architecture.
Assessed model performance by measuring detection accuracy, inference speed in frames per second (FPS), parameter count, and computational complexity (GFLOPs).
The LMDFS model achieved a detection accuracy of 80.2%, representing an absolute improvement of 5.9% over the baseline YOLOv7.
The architecture maintained a high inference throughput of 63.4 frames per second (FPS).
Model complexity was lowered by reducing overall parameter count by 14% and computational load by 6% GFLOPs.

Cite This Study

Gong et al. (2023) studied this question.

synapsesocial.com/papers/6a10d18439dd87f6d0ee6327 https://doi.org/10.3390/rs15153790

View Full Paper

Ask AI

Bookmark

Implication

Model evaluation study demonstrates improved smoke detection in aerial imagery, suggesting practical feasibility for early wildfire suppression via drones.

LMDFS: A Lightweight Model for Detecting Forest Fire Smoke in UAV Images Based on YOLOv7

Discussion

Member takes

Implication

Key Points

Cite This Study

LMDFS: A Lightweight Model for Detecting Forest Fire Smoke in UAV Images Based on YOLOv7

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider