Deep Learning in Satellite Imagery for Environmental Monitoring

Created using ChatSlide

This presentation delves into a pioneering approach in environmental monitoring using advanced deep learning techniques. Supervised by Ms. Ankita Nainwal and developed by authors Kush, Prakhar, Ananye, and Parth, the project introduces a multi-task learning model leveraging CNNs and Transformers to overcome the limitations of task-specific models. The methodology incorporates a U-Net framework with a ResNet50 encoder optimized using CrossEntropyLoss and the Adam optimiser. Results indicate...