Papers › A Real-time Fire Segmentation Method Based on A Deep Learning Approach

A Real-time Fire Segmentation Method Based on A Deep Learning Approach

22 Jul 2022IFAC-PapersOnLine 2022 7archive 2025-07-28

Mengna Li, Youmin Zhang, Lingxia Mu, Jing Xin, Ziquan Yu, Shangbin Jiao, Han Liu, Guo Xie, Yi Yingmin

As a kind of the forest “fault”, fire is highly destructive and difficult to rescue. Fire segmentation is helpful for firefighters to understand the fire scale and formulate a reasonable fire-fighting plan. Therefore, this paper proposes a real-time fire segmentation method based on deep learning. This method is an improved version of deeplbav3+, which is an encoder-decoder structure network. Encoder network is composed of deep convolutional neural network and atrous spatial pyramid pooling. Different from deeplabv3+, in order to improve the segmentation speed, this paper uses the lightweight network mobilenetv3 to build a new deep convolutional neural network and does not use atrous convolution, but it will affect the segmentation accuracy. Therefore, in order to compensate for the loss of segmentation accuracy, on the basis of the original decoder network, this paper adds two different shallow features to make the network contain rich fire feature information. Experimental results show that the comprehensive performance of this method is better than the original deeplabv3+, especially the segmentation speed of the network is greatly improved, which is about 59 FPS.

Paper PDF Code

Code

maidacundo/real-time-fire-segmentation-deep-learning pytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Decoder Real-Time Semantic Segmentation Segmentation

Results from the paper archive 2025-07-28

Task	Dataset	Model	Metric	Value	Rank at snapshot	Leaderboard	Report
Real-Time Semantic Segmentation	FLAME	Fast DeepLabV3+	FPS	59	#1 of 1	Archive leaderboard	report
Real-Time Semantic Segmentation	FLAME	Fast DeepLabV3+	Mean Intersection over Union	86.98	#1 of 1	Archive leaderboard	report
Real-Time Semantic Segmentation	FLAME	Fast DeepLabV3+	Mean Pixel Accuracy	92.46	#1 of 1	Archive leaderboard	report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

1x1 Convolution Average Pooling Batch Normalization Convolution Dense Connections Depthwise Convolution Depthwise Separable Convolution Dropout Global Average Pooling Hard Swish Inverted Residual Block Pointwise Convolution ReLU ReLU6 SPEED Sigmoid Activation Squeeze-and-Excitation Block

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections