Identification of steel surgical instruments using RGB-D orthoimage

In this study, we propose a simple method for identifying surgical steel instruments. This method detects instruments using Faster R-CNN applied to orthoimages acquired with an RGB-D camera and identifies them based on dimensional information. In the proposed method, an orthoimage is generated by orthographically transforming a 3D point cloud acquired using the Intel RealSense Depth Camera D435i. A Faster R-CNN object detector is applied to the orthoimages to detect surgical steel instruments. Since the generated orthoimages are constructed so that each pixel corresponds to a 1mm x 1mm area, dimensional information can be obtained directly from the image without requiring scale correction.