Object detection, a challenging task in computer vision, relies on models that are trained using real-world datasets. This research focuses specifically on object detection of baby products, and the ultimate goal is to strengthen Content-Based Image Retrieval (CBIR) systems, which have a wide range of applications including inventory management, AI-powered video analytics, product assembly, defect detection, and product improvement. The existing CBIR models primarily focus on adult products, neglecting the unique attributes of baby products. Forecasting baby product objects enables us to anticipate interactions with baby items, a crucial aspect for predicting actions involving infants. The You Only Look Once (YOLO) models are widely recognized for their precision and efficiency, making them suitable for various object detection applications. This study involves the development of a specialized dataset comprising baby products and the training of the YOLOv7-Tiny variant within the YOLOv7 framework. YOLOv7-Tiny is preferred because of experimental cost, and it was able to accurately predict small objects also. The experimental results for the proposed curated datasets called BP1 (Baby Products-1) and BP2 (Baby Products-2) reveal an average precision (AP) of 0.558 and 0.605, respectively.