Computer Vision (CV) is a study field that is responsible for developing techniques to perform tasks that the human visual system can do. Object detection is a technique used for detecting objects in videos and images. The research aims at detecting objects for indoor environment such as offices or rooms in different conditions of lighting by using YOLOv3 and generating a voice message for each detected object. YOLOv3 outperforms the other deep learning algorithms such as CNN because it looks at the entire image by predicting the bounding boxes using Convolutional Neural Network and finding class probabilities for these bounding boxes. However, CNN does not look at the image completely; it splits the image into regions that sequentially enter the neural network for performing the object detection and recognition process. This makes YOLOv3 faster than other deep learning algorithms. Open-source Computer Vision (OpenCV) was used for capturing the video frames. Then YOLOv3 was used to detect the objects in each frame and determine their location. Finally, the sound in the Arabic language was generated for the detected objects. The proposed method can detect six objects with an accuracy of 99% in the overall performance.