Real-time household object detection presents a formidable challenge due to the requirement for rapid computation in identifying objects promptly. Popular object detection models, such as Region-based Convolutional Neural Networks (R-CNN), Fast R-CNN, Faster R-CNN, You Only Look Once (YOLO), Spatial Pyramid Pooling Network (SPP-Net), Single Shot Multi-box Detection (SSD), and others, each come with their advantages and disadvantages. The present work proposes to use a Convolutional Neural Network (CNN) model known as SSD, as it processes images in a single step. Additionally, to address the computational constraints prevalent in devices such as mobile phones and laptops, a lightweight network model utilizing depth-wise separable convolution, termed MobileNet, is incorporated into the proposed methodology. The SSD-MobileNet model offers a mean Average Precision (mAP) of 21%, with the model weights ranging between 25 MegaBytes (MB), and achieves a processing speed of 30 milliseconds (ms) per image. Experimental findings demonstrate that integrating MobileNet with the SSD model significantly enhances accuracy in real-time household object identification. Furthermore, this research extends beyond theoretical exploration, showcasing the deployment of the SSD-MobileNet model within a flutter application, providing a user-friendly interface for seamless interaction with real-time detection outcomes.