Object detection and recognition via clustered features

The analysis of 2D images consists of two processes: detection and recognition of detected objects. Both stages allow for numerous applications in practical purposes, including detection of small objects and people with their appearance. The methods we can implement for these can benefit from fusion of approaches. In this article, we propose detection method based on analysis of the number of clusters of points in conjunction with Convolutional Neural Network as a final classifier. Proposed method of determining the clusters of points is based on a combination of modeled graphics processing with fuzzy logic. The proposed architecture of detection and classification has been tested and compared to other approaches in this field to show the efficiency and draw conclusions for further development.