Efficient Object Detection and Labeling in Retail Environments using MobileNetV2 with Inverted Residuals

Object detection and recognition are critical tasks in various applications, including retail, where accurate identification and labeling of objects can help manage inventory, track sales, and improve customer experience. Traditional methods for object detection and recognition are computationally expensive, require large amounts of training data, and may not generalize well to different environments and objects. To address these challenges, this research proposes a novel MobileNetV2-based deep learning model for identifying and labeling objects in a retail store environment. The proposed model is designed to be lightweight, efficient, and accurate, making it suitable for deployment on mobile devices or edge computing systems. The model is based on the Inverted Residuals Block (IRB) architecture, which allows for efficient use of computation and parameter resources while maintaining high accuracy. The proposed model is trained using a combination of image-level and object-level annotations and uses a multi-task loss function to optimize both the classification and localization performance. The experimental results show that the proposed model achieves high accuracy while minimizing computational resources required for deployment on mobile or edge devices. This research contributes to the development of an efficient and accurate machine learning model for object detection and recognition in retail environments.