An AI Driven Approach for Smoke and Haze Elimination Algorithm

The target of video dehazing is to reestablish fog free edges with amazing perceivability and difference in indoor fire mishaps through CCTV for the acknowledgment of the people who stuck under these fire risks and salvage them.Multi-scale Dehazing Combination (MDF) model with Profound Learning techniques like DCP (Dark Channel Prior) and TMAP (Transmission map) used to recuperate the corrupted cloudiness recordings by reproducing the video in high perceivability for tracking down the people in question.The model pre-processes the cloudy video into outlines.The foggy casings are then taken care of into the DCP and TMAP cycle to identify the smoke and barometrical light assessment from each edge.DMF model will encode and interpret each casing to manage debased recordings and Multi combination process supplies free data from one edge to another.A recreation module intended to remake high goal of cloudiness free video outlines.The result will help the salvage group to perceive individual through the web interface.This model was assessed on a dataset of 60000 murkiness outlines/pictures.The model accomplished with an exactness of 98% on testing with given input murkiness video which assists fire with safeguarding group to distinguish individuals who stuck under the fire situations.This could work on the precise identification of casualties in the indoor fire mishaps.