Vishal Batchu
I've been with Google Research since 2020 working on climate and sustainability. In the past, I have worked on quantization and binarization of deep learning models at IIIT Hyderabad focusing on the reduction of model sizes and improving their efficiency.
Research Areas
Authored Publications
Sort By
Mapping Farmed Landscapes from Remote Sensing
Alex Wilson
Michelangelo Conserva
Charlotte Stanton
CCAI Workshop at NeurIPS (2025)
Preview abstract
To overcome the critical lack of detailed ecological maps needed for managing agricultural landscapes, we developed Farmscapes: the first large-scale, high-resolution map that identifies ecologically vital rural features, including often overlooked elements like hedgerows and stone walls. We achieved high accuracy in mapping key habitats with a deep learning model trained on aerial imagery and expert annotations. As a result, this work enables data-driven planning for habitat restoration, supports the monitoring of key initiatives like the EU Biodiversity Strategy, and lays a foundation for advanced analysis of landscape connectivity.
View details
Satellite Sunroof: High-res Digital Surface Models and Roof Segmentation for Global Solar Mapping
Alex Wilson
Betty Peng
Carl Elkin
Umangi Jain
2024
Preview abstract
The transition to renewable energy sources such as solar is crucial for mitigating climate change. Google Maps Platform Solar API aims to accelerate this transition by providing accurate estimates of solar potential for buildings covered by aerial imagery. However, its impact is limited by geographical coverage and data availability. This paper presents an approach to expand the project's capabilities using satellite imagery, enabling global-scale solar potential assessment. We address challenges specific to satellite imagery, such as lower resolution and oblique views, by developing deep learning models for Digital Surface Model (DSM) estimation and roof plane segmentation. The models are trained and evaluated on datasets comprising of spatially aligned satellite and aerial imagery. Our results demonstrate the effectiveness of our approach in accurately predicting DSMs and roof segments from satellite imagery, paving the way for a significant expansion of the Solar API and impact in promoting solar adoption.
View details
Preview abstract
The need for characterizing global atmospheric carbon dioxide (CO2) concentrations at high spatial and temporal resolution is quickly increasing. Understanding the spatial and temporal variability of CO2 would provide the necessary constraints for quantifying anthropogenic CO2 emissions as well as natural carbon sources and sinks. Both of these are becoming increasingly important as the stability of natural sinks is increasingly uncertain and as the need to verify the effectiveness of climate mitigation efforts increases. The current generation of space-based sensors, however, can only provide relatively sparse observations in space and/or in time. While upcoming missions may address some of these challenges, most are still years away from launch. This challenge has fueled interest in the potential use of data from existing missions originally developed for other applications for inferring global greenhouse gas variability.
Sentinel-2 is a European wide-swath, high-resolution, multi-spectral imaging mission consisting of two satellites launched in 2015 and 2017 and may provide an important window into this problem given its high spatial resolution combined with its full spatial coverage over global lands (with a 5-day sun-synchronous revisit time). Sentinel-2 carries an optical instrument payload that samples 13 spectral bands including visible and near-infra-red (VNIR) and short wave infra-red (SWIR) bands.
Recent work has shown the potential for using data from Sentinel-2 to detect large point-source methane emissions. The use of vegetation indices derived from Sentinel-2 observations for monitoring terrestrial net ecosystem exchange and gross primary productivity have also been proposed. Here we explore the feasibility of deriving an atmospheric CO2 signal from Sentinel-2 observations. To do so, we train a U-Net styled deep neural network which takes all the Sentinel-2 bands from a short time series of Sentinel-2 imagery as the input and predicts a CO2 enhancement per pixel at 10m resolution. In order to train and validate the model, we generate synthetic CO2 plumes using an atmospheric model and embed these plumes into the Sentinel-2 image using Beer-Lambert radiative transfer at the current timestamp. The embedded CO2 plume enhancement in ppm is used as a target for the model. We also test our approach by looking at model predictions in regions where power plants are located and verifying plumes predicted by looking at chimney locations and wind velocities. In upcoming work, we will track improvements in the algorithm against observations from the Total Carbon Column Observing Network (TCCON), ACT-America and AirCore.
View details
Preview abstract
We develop a deep learning based convolutional-regression model that estimates the volumetric soil moisture content in the top ~5 cm. Input predictors include Sentinel-1 (active radar), Sentinel-2 (optical imagery), and SMAP (passive radar) as well as geophysical variables from SoilGrids and modelled soil moisture fields from GLDAS. The model was trained and evaluated on data from ~1300 in-situ sensors globally over the period 2015 - 2021 and obtained an average per-sensor correlation of 0.72 and ubRMSE of 0.0546. These results are benchmarked against 13 other soil moisture estimates at different locations, and an ablation study was used to identify important predictors.
View details
Cross-modal distillation for flood extent mapping
Shubhika Garg
Ben Feinstein
Shahar Timnat
Gideon Dror
Adi Gerzi Rosenthal
Environmental Data Science (2023)
Preview abstract
The increasing intensity and frequency of floods is one of the many consequences of our changing climate. In this work, we explore ML techniques that improve the flood detection module of an operational early flood warning system. Our method exploits an unlabelled dataset of paired multi-spectral and Synthetic Aperture Radar (SAR) imagery to reduce the labeling requirements of a purely supervised learning method. Prior works have used unlabelled data by creating weak labels out of them. However, from our experiments we noticed that such a model still ends up learning the label mistakes in those weak labels. Motivated by knowledge distillation and semi supervised learning, we explore the use of a teacher to train a student with the help of a small hand labelled dataset and a large unlabelled dataset. Unlike the conventional self distillation setup, we propose a cross modal distillation framework that transfers supervision from a teacher trained on richer modality (multi-spectral images) to a student model trained on SAR imagery. The trained models are then tested on the Sen1Floods11 dataset. Our model outperforms the Sen1Floods11 baseline model trained on the weak labeled SAR imagery by an absolute margin of 6.53% Intersection-over-Union (IoU) on the test split.
View details
Cross Modal Distillation for Flood Extent Mapping
Shubhika Garg
Ben Feinstein
Shahar Timnat
Gideon Dror
Adi Gerzi Rosenthal
Tackling Climate Change with Machine Learning, NeurIPS 2022 Workshop
Preview abstract
The increasing intensity and frequency of floods is one of the many consequences of our changing climate. In this work, we explore ML techniques that improve the flood detection module of an operational early flood warning system. Our method exploits an unlabelled dataset of paired multi-spectral and Synthetic Aperture Radar (SAR) imagery to reduce the labeling requirements of a purely supervised learning method. Past attempts have used such unlabelled data by creating weak labels out of them, but end up learning the label mistakes in those weak labels. Motivated by knowledge distillation and semi supervised learning, we explore the use of a teacher to train a student with the help of a small hand labeled dataset and a large unlabelled dataset. Unlike the conventional self distillation setup, we propose a cross modal distillation framework that transfers supervision from a teacher trained on richer modality (multi-spectral images) to a student model trained on SAR imagery. The trained models are then tested on the Sen1Floods11 dataset. Our model outperforms the Sen1Floods11 SAR baselines by an absolute margin of 4.15% pixel wise Intersection-over-Union (IoU) on the test split.
View details
