BhoomiDrishti : View of the Land

Inspiration

We were inspired to work on this project by the growing importance of satellite imagery in solving real-world problems such as deforestation, urban growth, and climate change. The idea of combining geospatial data with deep learning and generative AI excited us because it could not only help in monitoring land use and environmental changes but also create new ways of storytelling about places and their cultural significance.

Learnings

Through this project, we came to know about how to deal with geospatial data, and understood the importance and impact of geospatial data in various applications.

Challenges

The main challenges we might face are the large size of satellite datasets, cloud cover issues in imagery, and limited labeled data for training. Issues of time and space complexity (e.g., $O(n^2)$ for image processing in some cases) also arise. Training deep learning models on such high-resolution data requires high computational resources, and ensuring that the outputs are both accurate and interpretable is a challenge. Generative models, while powerful, often produce unrealistic or inconsistent features, so validating results carefully is necessary.

Built With

• classification
• deep-learning
• llms/gan
• technologies-python