Surgical-Tools Tracking Based on Convolutional Neural Network and Long Short-Term Memory

Real-time surgical tool tracking is a critical component of computer-assisted surgery, because it is highly instrumental to analyze and understand the surgical activities. Nowadays, many deep learning methods take fully advantage of very deep networks and track by detection. Although these methods work well, but they take up a significant amount of time and computational resources. To address this problem, we propose a new network which use the cascade of refined convolutional neural network and long short-term memory for real-time single tool tracking based on the Real-time Recurrent Regression Networks (Re3). Our method is tested on the publicly available standard dataset from UCL (University College London). The experimental result show that our method achieves better performance than state-of-the-art tracking methods in terms of accuracy and speed.