Multiclass Object Detection for Surgical Instruments using YOLOv3

Ayush Shetty     

ayush.shetty@duke.edu    

Paper PDF

Before any major surgery, it is the nurse�s job to prepare all the important surgical instruments and setup the tray. All instruments must be accounted for immediately after the surgery so that no instrument is left behind inside the patient. Cases of emergency or human error post or mid surgery might cost very vital minutes, which could lead to severe complications or even patient death in a worst-case scenario. This project aims to help the nurses effectively and completely arrange the instrument tray before and after the surgery to account for all the instruments used in the surgery using computer vision.

Paper: Paper PDF

Code and Data: Code: Code Data set: Kaggle dataset