Automated monitoring systems have proven to be effective in improving the productivity of equipment-intensive operations in the construction and mining sectors. Vision-based systems are the most recent methods employed to detect, track, and monitor construction equipment. The currently developed systems, however, analyze video frames captured by a stationary camera, which dramatically limits their coverage area and requires manual adjustment of the viewfinder. This research paper introduces methods to proactively steer a pan-tilt-zoom camera to localize, track, and identify objects of interest in construction jobsites. This automated camera control system uses a number of image and video processing algorithms to detect objects and estimate their trajectory and velocity, and then uses the extracted information to set the camera movement parameters, including direction and magnitude. The experimental results of this system showed promising performance for equipment monitoring in construction and mining jobsites.