The Use of Computer Vision Methods to Detect People in Forest-Steppe Terrain

The approach of using learning methods for adjustment of neural network models used for automatic detection of people in the forest-steppe areas in images was studied. For the detection, the neural network models YOLOv51, DenseNet, CenterNet, InceptionV3, Xception, and Faster R-CNN were used. Among the presented models, the best results were demonstrated by the neural network YOLOv51 subjected to fine-tuning, which allowed identifying people by graphic images with an accuracy of 0.8795 using the test sample according to the F1-score metric with a threshold value of 0.5.