In this study, we detect and locate the disinfection objects and determine the scope of the disinfection in public places. Firstly, a depth camera was used to capture RGB images and a three-dimensional (3D) point cloud of the public disinfection objects. Secondly, using trained Mask R-CNN, the classification, detection, and instance segmentation of disinfection targets are carried out, yielding a 3D point cloud of the objects. The 3D point clouds from different perspectives were then stitched together to create a complete 3D point cloud of the disinfection object using the sample consensus initial alignment (SAC-IA) and iterative closest point (ICP) fine registration methods. Finally, the 3D point cloud's bounding box was optimized using principal component analysis (PCA). The experimental results show that the mean average precision(mAP) of object detection based on Mask R-CNN reaches 0.968, and the average intersection over union (IoU) of instance segmentation reaches 0.879. The optimization rates of surface area and volume of the bounding box are 29.2% and 28.8%, respectively. The effectiveness of this method in detecting and locating disinfection objects was demonstrated in this study. It lays a foundation for providing different disinfection methods suited for different objects, narrows the disinfection scope while simultaneously improving disinfection efficiency.