Grain Wagon Fill Detection using Camera and Deep Convolution Network

摘要

The article provides a description of the system of grain stream automatic control when grain is loaded into the grain box. The system uses the data from only one camera and the deep convolutional network for the box elements and grain stream recognition. A rigid fixation of the camera above the grain box allows to use the bird's-eye viewmodel of transformation for generating signals about the location of the box relative to the auger pipe.