Real-time infield weed detection is the most important technique of the precision infield weeds control. The real-time weed detection equipment is a key of the site specific weed management. Currently research is being carried out relating to weed sensor based on the spectroscopic, imaging and spectral imaging techniques. Theses advance technologies and equipments were reviewed for application them in China. The weed detection technique based on spectroscopic is suitable for classifying weeds from soil background without crop seedlings, such as the WeedSeeker and Weed-IT. The weed detection technique based on imaging is appropriate for discriminating inter-row weeds, including the Robocrop InRow mechanical weed control machine and the vision navigation product of the Autopilot, Cam Pilot and Robocrop, etc. The weed detection technique based on spectral imaging is more appropriate for distinguishing intra-row weeds, and a joint research on microphotonic plant discrimination and weed control sensor has been done by China and Australia. It is still a continuous project that how to detect infield weeds in real-time and specific under the complex and open field environment.