The implementation showcased in this repository revolves around employing deep learning models for the surveillance of disasters using drone-based imagery. The primary focus involves developing models for two essential tasks: Image Inpainting, utilizing the c-GAN architecture, and Image Classification, with a specific emphasis on disaster-related images. The approach for Image Inpainting involves the c-GAN architecture, while Image Classification is executed through fine-tuning ResNet-50.
Students enrolled in the CS F425 - Deep Learning course in the first semester of the academic year 2023-24 curated the training and validation images. These images were sourced from a range of drone-based videos available on the internet.
For any inquiries or questions regarding the model implementation, feel free to contact the project maintainer at [email protected]