Researchers used climate data and satellite technology to develop the predictive model
CHANDIGARH, India, June 27, 2026 /PRNewswire/ — In a significant advance in smart farming and precision farming, researchers from the University of Chandigarh have developed an Artificial Intelligence (AI)-powered Transformer Model that is capable of accurately predicting crop yields using satellite images, climate data and historical agricultural data.
The innovation would enable farmers, policymakers and agricultural agencies to make informed decisions while strengthening food security and promoting resilient agricultural management.
The research, led by Kusum Lata, Assistant Professor, Department of Computer Science, Chandigarh University, Dr. Navneet Kaur, Professor, Department of CSE and Dr. Simrandeep Singh Professor, University Center for Research and Development, Chandigarh University, focuses on improving crop yield prediction in the agricultural heartland of Punjab. The study, recently presented at the 2026 International Conference on Signal Processing and Electronics Design (ICSPED) at Chandigarh College of Engineering and Technology, Chandigarh, introduces a lightweight transformer-based system that uses data from multiple sources to estimate pre-harvest crop production with higher accuracy and lower computational costs.
In particular, accurate crop yield prediction has become increasingly important as farmers face increasing challenges due to climate variability, changing weather patterns and rising food demand. Traditional field studies are often time-consuming, labor-intensive and limited in scope. The researchers from Chandigarh University sought to overcome these limitations by integrating advanced AI techniques with real-time Earth observation data.
Kusum Lata, Assistant Professor of Computer Science Engineering at CU, said: “The Transformer model uses data from Sentinel-1 and Sentinel-2 satellites. These are advanced Earth observation satellites operated by the European Space Agency (ESA) to continuously monitor agricultural fields and provide information on crop growth, vegetation health, soil moisture and field conditions. The satellite observations are combined with climate variables such as rainfall, temperature and soil moisture, along with historical crop production data, providing a comprehensive picture of crop performance throughout the growing season.”
Kusum added: “Unlike conventional machine learning models, the newly developed lightweight transformer can identify critical growth stages of crops and learn complex temporal patterns that influence the final yield. We designed the model to deliver high predictive performance that requires fewer computing resources, making it suitable for practical deployment in large-scale agricultural monitoring systems.”
“The model was evaluated on four major crops grown in Ludhiana district, namely rice, maize, moong and sugarcane, using data collected between 2019 and 2023. Experimental results showed that the transformer model outperformed commonly used approaches such as Random Forest and Long Short-Term Memory (LSTM) models, indicating a stronger match between predicted and actual yields. The framework also recorded reduced prediction errors and improved computational efficiency.
Kusum has worked as a Junior Research Fellow at the Agriculture and Crop Monitoring Division of the Punjab Remote Sensing Center (PRSC), PAU Ludhiana, and has contributed to geospatial research projects including analysis of crop residue burning using remote sensing and geospatial mapping techniques.
The study further found that the lightweight architecture requires almost 40 percent fewer parameters than conventional transformer models, while delivering faster and more accurate predictions. This level of efficiency makes this automated framework suitable for near real-time agricultural applications, including regional crop monitoring, production forecasting and early warning systems.
According to the researchers, the ability to accurately predict crop yields before harvest could have significant implications for both farmers and governments. Reliable forecasts can support agricultural planning, optimize resource allocation, strengthen crop insurance mechanisms and improve market management strategies. In a state like Punjab, where agriculture plays a central role in the economy, such technologies can contribute to more resilient and sustainable agricultural systems.
The researchers also shared that one of the model’s key strengths lies in its ability to combine multiple sources of information into a single predictive model. By integrating satellite observations with climatological and historical datasets, the system captures the complex interactions that influence crop productivity and provides a more robust understanding of agricultural outcomes.
Future developments will also focus on enabling near real-time predictions through cloud-based platforms, paving the way for wider adoption of AI-driven decision support systems in agriculture, the Chandigarh University researchers added.
About Chandigarh University
Chandigarh University is a NAAC A+ Grade University and QS World Rated University. This autonomous educational institution is recognized by UGC and is located near Chandigarh in the state of Punjab. It is the youngest university in India and the only private university in Punjab to be honored with an A+ grade by NAAC (National Assessment and Accreditation Council). CU offers more than 109 UG and PG programs in engineering, management, pharmacy, law, architecture, journalism, animation, hotel management, commerce and others. It has been awarded by WCRC as The University with Best Placements.
Website address: https://www.cuchd.in/
Photo: https://web3wire.org/wp-content/uploads/2026/06/Chandigarh_University_AI_Crop_Yield.jpg
View original content to download multimedia:https://www.prnewswire.com/in/news-releases/chandigarh-university-researchers-use-artificial-intelligence-to-develop-model-for-predicting-accurate-crop-yield-innovation-to-benefit-indian-farmers-302812564.html