Aquaculture Pond Precise Detection and Monitoring for Spacial Planning Using Deep Learning and Remote Sensing

The research note introduces a deep learning-based framework using Sentinel-2 imagery to identify, classify, and monitor aquaculture ponds across India’s coasts. By integrating NDWI-based segmentation, DeepLabv3 architecture, and Random Forest classifiers, the system achieves precise boundary mapping and area estimation. It further tracks 30-year spatiotemporal changes, helping policymakers and researchers promote sustainable aquaculture development and better marine resource management.

1. Deep Learning Integration: Combines NDWI, DeepLabv3, and Random Forest for high-accuracy pond detection and classification.

2. Automation & Accuracy: Enables large-scale, automated monitoring with minimal human input and strong performance across complex pond shapes.

3. Environmental Insight: Tracks long-term spatial changes to assess impacts of urbanization, pollution, and climate change on aquaculture zones.