An Automatic Drill Pipe Counting Method for Gas Drainage Based on Improved SiamRPN

Gas drainage is one of the most effective ways to prevent the risk of gas explosion. To solve the problems of low efficiency and dependability of human experience by manual counting the number of drill pipes for gas drainage, this paper proposes an automatic counting framework for borehole depth measurement based on visual tracking. Firstly, a path-constrained approach is adopted to improve the Siamese network for real-time drilling rig tracking, addressing the problem of frequent tracking loss caused by occlusion and exposure. Then, the reciprocating motion pattern of the drilling rig is analyzed, and a parameteradaptive pipe counting scheme is proposed for complex drilling scenarios, in which the number of reciprocating motions of drilling rig for extracting a drill pipe varies. Through experiments on 20 videos collected from underground sites using different drilling equipment, our proposed scheme achieves a high accuracy rate of 99.5% for drill pipe counting in complex motion scenarios, showing excellent robustness.