Optimization of Vehicle Object Detection Based on UAV Dataset: CNN Model and Darknet Algorithm

This study was conducted to identify several types of vehicles taken using drone technology or Unmanned Aerial Vehicles (UAV). The introduction from above an altitude more than 300-400 meters that pass the highway ground level becomes a problem needs optimum investigation so there are no errors in determining type vehicle. at mining sites class through and how many road for vehicle recognition deep learning algorithm CNN models such as Yolo V4, V3, Densenet 201, CsResNext –Panet 50 supported by Darknet support training process. In this study, experiments were carried out with other models, but peripherals hardware devices, only 4 resulted optimal accuracy. Based on experimental results, CSResNext-Panet model has highest accuracy can detect 100% captured UAV video data, including number detected volumes, then which up 98% - 99%. research continue be developed knowing all classes affordable must peripheral