Published Paper Details:

Image recognition, a crucial facet of computer vision, involves the automated identification and categorization of visual content within images, playing a pivotal role in diverse applications such as medical diagnostics, autonomous vehicles, security systems, and augmented reality, significantly enhancing efficiency and accuracy in various domains. This study explores and compares various object detection and classification methods, incorporating LBP,Haar Cascade, HOG for detection, and for classification CNN, DNN. Hybrid methodologies, including Haar Cascade with CNN, Haar Cascade with DNN, LBP with CNN, LBP with DNN, HOG with CNN, HOG with DNN, were rigorously tested on different embedded systems utilizing the Microsoft COCO dataset. Results revealed that the Haar Cascade with CNN methodachieved the highest recognition success rate at 78.60%, surpassing other methods. These outcomes highlight the efficacy of the Haar Cascade with CNN approach, especially on powerful embedded systems, showcasing its potential for real-time object recognition applications