Deep Learning

Summary: Notes on deep learning architectures, models, and applications in geospatial and other domains. Last updated: 2026-08-20

---

2026

• Antelope Valley Wildflower Forecast: An AI-powered bloom prediction tool for California poppies and goldfields in Antelope Valley, using deep learning on satellite imagery and weather data (accumulated precipitation, growing degree days, wind speed) to forecast wildflower blooms 5 days in advance. Related: Remote_Sensing, Climate, AI. Keywords: deep learning, remote sensing, wildflower, bloom prediction, satellite imagery, weather, California

• Urban Change Forecasting from Satellite Images: Uses a UNet model to predict imminent urban change from satellite imagery, enabling proactive urban planning and monitoring. Related: Remote_Sensing, Urban_Planning, Climate. Keywords: deep learning, urban change, UNet, satellite imagery, urban planning, forecasting

2025

• GeoDeep: A lightweight Python library for AI-powered object detection and semantic segmentation on geospatial raster images, with pre-trained models included for immediate use. Related: Python, AI. Keywords: deep learning, geospatial, object detection, segmentation, Python, pre-trained

• Geo Deep - Sentinel-2 Super Resolution (Colab): Automated Sentinel-2 deep learning super-resolution inference notebook for upscaling 10m to 5m resolution. Related: Remote_Sensing, Python. Keywords: deep learning, Sentinel-2, super resolution, Python, Colab

Earlier

• satellite-image-deep-learning: Comprehensive resource list for performing deep learning on satellite imagery. Related: Remote_Sensing. Keywords: deep learning, satellite imagery, resources, remote sensing

• Crop Classification CNN Fastai: Crop classification using convolutional neural networks with Fastai framework. Related: Agriculture. Keywords: deep learning, crop classification, CNN, Fastai, agriculture

• Development Seed TensorFlow EO Tutorial: Development Seed tutorial series on using TensorFlow for Earth observation tasks. Related: Remote_Sensing, Python. Keywords: deep learning, TensorFlow, Earth observation, remote sensing, Python

• Hill Shade with Neural Networks: Paper on cartographic relief shading using neural networks. Video Related: Cartography. Keywords: deep learning, cartography, hillshade, neural network, visualization

• Vessel Tracker - Ship Counting with Sentinel-1 and GEE: Ship counting and tracking using Sentinel-1 SAR imagery and Google Earth Engine. Related: Remote_Sensing, GEE. Keywords: deep learning, Sentinel-1, SAR, ship detection, GEE

• WorldStrat Super Resolution SPOT: Open-source high-resolution satellite imagery super-resolution using SPOT imagery. Related: Remote_Sensing. Keywords: deep learning, super resolution, SPOT, satellite imagery

• Transformers Hugging Face: Hugging Face repository of transformer models for NLP and vision tasks applicable to Earth observation. Related: AI. Keywords: deep learning, transformers, Hugging Face, NLP, vision

• CloudSen12 - Cloud Detection for Sentinel: Deep learning dataset and models for cloud detection in Sentinel-1 and Sentinel-2 imagery. Related: Remote_Sensing, Data. Keywords: deep learning, cloud detection, Sentinel-2, dataset

• Satellite Image Time Series Annotated Datasets: Collection of annotated datasets for satellite image time series deep learning research. Related: XArray, Remote_Sensing, Data. Keywords: deep learning, time series, satellite imagery, dataset, annotation

• Satellite Image Normalization for Deep Learning: Sentinel Hub guide on how to properly normalize satellite images for deep learning applications. Related: Remote_Sensing. Keywords: deep learning, satellite imagery, normalization, preprocessing, Sentinel Hub

• DeepWater: Deep learning model for global surface water mapping from satellite imagery. Related: Remote_Sensing. Keywords: deep learning, water mapping, global, satellite imagery