The present invention is a method for automatically configuring a system for recognizing a class of variable morphological objects, comprising an initial data set sufficient to recognize instances of the class object in a sequence of images of the scene of interest. Providing a machine learning system having (10), providing a comprehensive three-dimensional model specific to a class of objects whose morphology can be defined by a set of parameters; and using a camera (12). Acquiring a series of images of the scene; recognizing an image instance (14) of an object of a class in the acquired series of images using the initial data set; (16) Step to fit And storing a range of parameter variations resulting from the fitting of the generic model (20); and synthesizing a plurality of three-dimensional objects from the generic model by varying the parameters within the stored variation range (20). 22) and supplementing the learning system dataset (10) with the projection (24) of the synthesized object in the plane of the image. [Selection diagram] FIG.
展开▼
