Identification of objects is considered as crucial and important task in computer vision field, which is widely employed for numerous applications. Essentially, identification of objects methods merges classifying as well as localizing objects for the purpose of predicting both single and multi-objects in the acquired pictures. UAVs (Unmanned Aerial Vehicle) images are documenting with variant scales or resolution based on the flight altitude of UAVs. The detailed description of UAVs images objects might get missed because of the smaller dimension and size of these images, which results an insufficient functioning of detection of framework which are small in nature. The small objects are suffered from poor visual data. They are also considerably vulnerable to environmental fluctuations, making it challenging for detection methods for the purpose of precisely locating as well as classify small objects in UAVs images. For handling the defined problem statement, Yolov8 deep learning model is utilized in this paper to detect five types of small objects in UAVs images and pre-trained on custom database. Furthermore, database preparing and fine alignment techniques of the main hyperparameters are achieved in the proposed framework to make deep learning model more fit to the training data. Evaluation of effectiveness of the trained mechanism is achieved using VisDrone2019 database which compressed variant images captured through Unmanned Aerial Vehicles in various categories such as climate setting, resolution along with altitude. The finding exhibited the outperformance of the proposed framework and getting 42.7% mAP metric compared to the state of art approaches-based small identification of objects task.