Strip Pooling Coordinate Attention with Directional Learning for Intelligent Fire Recognition in Smart Cities

Fire detection in smart cities requires intelligent visual recognition systems capable of distinguishing fire from visually similar phenomena while maintaining real-time performance under diverse environmental conditions. Existing deep learning approaches employ attention mechanisms that aggregate spatial information isotropically, failing to capture the inherently directional characteristics of fire and smoke patterns. This paper presents DirFireNet, a novel fire detection framework that exploits directional fire dynamics through Strip Pooling Coordinate Attention (SPCA). Unlike conventional attention mechanisms, DirFireNet explicitly models vertical flame propagation and horizontal smoke dispersion via directional strip pooling operations that decompose features along horizontal and vertical axes. The framework integrates a progressive top-down fusion pathway with attention-guided weighting that synthesizes multi-scale representations from coarse to fine resolutions. Furthermore, dual global pooling captures complementary scene statistics holistic fire intensity and salient flame regions. Built upon the lightweight EfficientNetV2-S backbone, DirFireNet achieves superior accuracy while maintaining computational efficiency. Extensive experiments on the FD and BoWFire benchmark demonstrate state-of-the-art (SOTA) performance. Comprehensive ablation studies validate that directional attention contributes to accuracy gain, validating that attention mechanism provides strong inductive biases for intelligent fire recognition in smart city applications.