This repository focuses on detecting vehicles from satellite imagery using state-of-the-art (SOTA) object detection models and software. Leveraging cutting-edge techniques, the project aims to achieve high accuracy in identifying and classifying vehicles within complex satellite images.
This repository focuses on detecting vehicles from satellite imagery which can be can be aquired using QGIS followed by using state-of-the-art (SOTA) object detection models and software. Leveraging cutting-edge techniques, the project aims to achieve high accuracy in identifying and classifying vehicles within complex satellite images.
- SOTA Object Detection Models: Utilizes advanced object detection models like YOLOv5, YOLOv10, and others, known for their superior performance in various object detection tasks.
- Image Augmentation: Implements sophisticated image augmentation techniques to enhance model robustness and generalization.
- Hyperparameter Tuning: Employs systematic hyperparameter tuning strategies to optimize model performance, ensuring the best trade-off between precision, recall, and mean Average Precision (mAP).
- Visualization Tools: Includes tools for visualizing detections and performance metrics, providing clear insights into model performance and areas for improvement.
- TensorFlow and PyTorch: Primary deep learning frameworks used for model development and training.
- OGIS : Used to load up Google satellite .tiff files and prepare a rastor followed by exporting geotagged data of the grids.
- Google Colab: Employed for running experiments and training models with GPU support.
We welcome contributions from the community. Please feel free to submit issues, fork the repository, and send pull requests.
This project is licensed under the MIT License. See the LICENSE file for more details.