Comparative analysis of CNN-based smart pre‐trained models for object detection on DOTA

In this paper, we proposed a comparative research project on the classification of various objects in satellite images using some pre-trained models of CNN (VGG- 19, ResNet-50, Inception-V3, EfficientNet-B7) and R-CNN. In this research work, we have used the DOTA dataset, which combines data from 14 classes. We have imple- mented above-mentioned pre-trained models of CNN and R-CNN to achieve optimal results for accuracy as well as productivity in detection of various objects such as ships, tennis courts, swimming pools, vehicles, and harbors from remotely accessed images. In this study, a convolutional neural network (CNN) is used as the base model. For complex computations and for speeding up results, transfer learning is used. With the help of experimental analysis, we have discovered that R-CNN and Inception-V3 performed best out of the five pre-trained models

---

Opracowanie rekordu ze środków MNiSW, umowa nr POPUL/SP/0154/2024/02 w ramach programu "Społeczna odpowiedzialność nauki II" - moduł: Popularyzacja nauki (2025).