Abstract: This article selected and used leaves of a common ornamental plant Euonymus japonicus as detected nursery stock, based on the OpenCV, wroe a program which could detect and identify E. japonicus leaf, through training, Boosted classifier to detect E. japonicus was obtained, and by detecting the Haar characteristics, found out a dynamic detection method which could detect the E. japonicus leaves from complex background video image, be satisfied with the requirements of the detection and identification of weeding in nursery. Confirmed by test data and the actual test results, this object detection method is feasible and convenient.