Deep learning has developed into a potent machine learning technique that combines several feature layers or data models to provide state-of-the-art outcomes. In many areas, such as object detection, the process of segmentation and image classification, deep learning has proven to function exceptionally well. In recent years, there has been a notable advancement in fine-grained picture classification using deep learning algorithms, whereby aim to distinguish between classifications at a lower level. This task has modest inter-class variance and considerable intra-class variance. In self-driving vehicle applications, identifying objects and pedestrian recognition are essential. Techniques built around convolutional neural networks have demonstrated notable improvements in precision and decision efficiency in real-time applications. The researchers of this study describe a number of cutting-edge deep learning methods, including InceptionV3, VGG-16, VGG-19, DenseNet-121, and a specially designed three-layer CNN system for object detection. The chosen model is trained and verified using a dataset of five classes of furniture that the user created. Following thorough testing, VGG-19 produced the maximum accuracy of 99.89%.