Object Recognition in Images Using Hybrid Deep Learning Model

Abstract

Abstract: Classifying and identifying objects through images and making bounding boxes is the basicobjective of object recognition and detection.Object recognition, is the most crucial problem,this being the reason it hasreceived a strong attention for the research. With the huge growth of object detection technology in computer vision over the last few years, the subject has seen a significant change. In the 1990s, people were still using creative thought and long-lasting design to figure out how to recognize objects in early computer vision. If you look at how we identify objects today as a change made possible by deep learning, you can learn both high-level and low-level features. This paper discusses blended approach in the field of object recognition through deep learning. Major contribution of this work is to present a hybrid classifier approach with some of prominent backbone architecture using EfficientNet CNN Deep learning model combined with YOLO detector for the object recognition named E-YOLO.On some metrics this model test with some existing model on MS COCO dataset for the Common benchmark. Lastly comparison of the performance and accuracy of existing model with proposed model on these metrics has been discussed. As a result the accuracy of proposed model is better than the existing model.