Indoor fire hazards pose critical challenges due to obscured vision caused by thick smoke, impeding firefighters’ navigation and decision-making abilities in real time. Existing methods often fall short of addressing these dynamic scenarios effectively. This paper introduces an innovative AI-ML-based algorithm for intelligent de-smoking/dehazing to mitigate this challenge. Central to the system is a real-time dehazing framework leveraged by the Feature Fusion Attention Network (FFA-Net), tailored for image dehazing within indoor fire environments. FFA-Net integrates feature attention, local residual learning, and attention-based feature fusion, enhancing visibility in real-time video streams. Additionally, a mobile application interfaces seamlessly with FFA-Net, providing firefighters with a user-friendly platform to access dehazed video feeds. This advancement in visual enhancement fosters improved situational awareness and informed decision-making during critical rescue operations amidst indoor fire emergencies.