A Hybrid Deep Learning Approach for Accurate and Efficient Object DetectionOpen Access

Abstract

Humans can easily identify multiple objects in an image or video but for computers, it is very difficult to identify. The procedure of precisely detecting and promptly recognizing items is quite challenging. However, with the assistance of diverse object detection algorithms, we can now do this work with utmost precision. The proposed method achieved an accuracy of 0.78 to 0.84 on various objects. The advantage of embedding mask RCNN and yolo v7 was achieved with a good precision value. The experimental results were published by masking the specific object and masking the background of the image. The Advantage of using Embedded with Mask R-CNN and YOLO v7 is that it achieves a good precision value. The experimental results were published with masking applied to a specific object and its background. The proposed method concluded that, with YOLO v7, we could reduce the computational effort by 30% and parameter optimization by 40% compared to the existing method.