A Lightweight Algorithm for Detecting Smoke in Forests without Open Flames

Fast and accurate judgment of forest fire is of great significance to forest fire prevention. Most of the existing forest smoke detection models are only applicable to the case where there is an open fire in the smoke image, and the excessive model volume makes it difficult to be applied to edge devices. To address this problem, a lightweight forest smoke detection algorithm without open fire is proposed. The algorithm introduces the attention mechanism CA and the full convolutional mask self-encoder framework FCMAE in the backbone network, so that the model can efficiently extract semantic information and high and low level features while solving the feature collapse problem of existing models. A centralized feature pyramid CFP is also introduced in the prediction network to enhance the intra-layer conditioning of features. In addition, the model uses the loss function Wise-IoU with dynamic non-monotonic FM to strengthen the detection ability of low-quality smoke samples. Experimental results show that the model has the best performance in detecting smoke without open flame compared to other lightweight models.