Dynamic Inventory Object Detection Using YOLOv8 Deep Learning Model

Inventory object detection is the process of locating and recognising items inside the warehouse or inventory using deep learning and computer vision techniques. Computer vision, which employs object detection, is utilized for classifying and identifying objects within images. This method is applied in various fields, like inventory, to help in identification of various objects. Among business sectors, one of the greatest obstacles still lies in effectively managing inventories in dynamic environments. In this paper, this work presents the development and optimization of the You-Only-Look-Once, YOLOv8, object detection technique to identify various object types and their locations in an inventory context. This approach ensures adaptability and robustness by utilizing transfer learning on pretrained models, which allows it to easily adjust to different inventory scenarios. For dependable performance in dynamic environments, a novel approach is suggested for enhancing data that will improve generalizing the model and reduce overfitting. On an annotated dataset of inventory objects, the model is trained and classifies them as objects or not. This paper highlights the improvement of identification of relevant objects in an area of inventory. The adoption of this model improves the result by identifying the items as object or not.