Water Resources and Power

Hydraulic engineering generally has large earthmoving demand,and dump-trucks are the main delivery machinery.Statistics of truck earthmoving quantity is important for project management.Manual ledger or weighbridge are currently the main approaches for truck earthmoving quantity statistics,which have shortcomings of human error,high cost,low efficiency and limited application.This paper proposes a machine vision based non-contact method for intelligent truck earthmoving quantity statistics.The method can recognize the loading conditions of empty/full for earthmoving trucks through site surveillance video images,and obtain the truck information with rapid sensing technology and database to achieve intelligent statistics.The method does not interfere earthmoving activities,is labor-free,and utilizes the field surveillance system,thus it possesses advantages to current methods in cost and efficiency.Case study shows that the proposed method can achieve non-contact intelligent truck earthmoving quantity statistics on field,and its deep-learned machine vision model has fast speed and good accuracy in recognizing empty/full loaded trucks.