Nowadays Drones are used in various places, which increases the advancement in the innovation of drones. According to that drones were modified with various design segments. In general, it is used for security and surveillance. In this article, modified YOLOv5 network with an additional layer neck -PANet is presented. It is presented to recognize four types of drones (multirotor, fixed-wing, helicopters, and Vertical Take-Off and Landing (VTOLs) to differentiate them from birds with a set of 1000 visible images. In this network, more effective and detailed semantic features were extracted by changing the number of convolutional multi-layers. The performance of the basic YOLOv5 network was also evaluated on the same dataset. The proposed model performs well 5% with the existing art of YOLOv3, YOLOv4 basic models. In YOLOv5 will be exponentially more comfortable for object detection.