Aquila Optimization Driven Deep Learning Framework for Emergency Vehicle Alerting and Recognition in Dense Indian Traffic

Emergency Vehicles (EV) in congested traffic is
important for assuring rapid response and reducing important
interruptions an effective appreciation and prioritizing. For accurate and real-time identification the structure uses image processing and Deep Learning (DL) algorithms. The raw input data is fed to pre-process with an Adaptive Bilateral Filter (ABF), which efficiently reduces noise whereas protecting critical edge features. The improved images are then fed toFuzzy Cluster-Based Thresholding (FCBT) for precise segmentation of EV in contrast to difficult traffic backgrounds. To demonstrate discriminative visual patterns, Discrete Wavelet Transformation (DWT) is utilized as a multi-level feature extraction. The extracted features are then recognized using a unique Deep Attention Dilated Residual Aquila Convolutional Neural Network (DADR-AquilaNet), which combines dilated convolutions, residual learning, and attention processes to improve feature sensitivity as well as contextual awareness.
Further to fine tune DADR, Aquila Optimization Algorithm
(AOA) is to regulate the network limits, resulting in better
convergence and classification resilience. An experimental
evaluation is implemented by Indian Emergency Vehicles
Dataset displayed an enhanced detection accuracy of 98% and
occlusion situations. The structure offers an intelligent and scalable solution for real-time EV alerts in congested urban traffic.