Model and training method of moving object classification system for a compact drone

Abstract

The classification model which consist of motion detector, object tracker, convolutional sparse coded feature extractor and stacked information-extreme classifier is developed. Proposed model is characterized by low computational complexity and it can be used as labeling dataset gathering tool for deep moveable object detector. Furthermore, training method for moving object detector is developed. The method consisting in unsupervised pre-training feature extractor based on sparse coding neural gas, supervised pre-training and following fine-tuning of stacked information-extreme classifier. Labeling new emerging data through self-labeling for high prediction score cases and manual labeling for low prediction score cases, and following labeled object tracking are also offered. In this case, class balancing using undersampling within dichotomous strategy “one-against-all”. Simulation results on open datasets confirm the suitability of proposed model and training method for practical usage.