MAgSeg: Segmentation of Agricultural Landscapes in High-Resolution Satellite Imagery using Multimodal Large Language Models

Tiwary, Piyush; Ahuja, Utkarsh; Sani, Depanshu; Jayagopal, Aishwarya; Gubbi, Sagar; Venugopalan, Subhashini; Talekar, Alok; Rajan, Vaibhav

doi:10.24963/ijcai.2026/728

MAgSeg: Segmentation of Agricultural Landscapes in High-Resolution Satellite Imagery using Multimodal Large Language Models

Piyush Tiwary, Utkarsh Ahuja, Depanshu Sani, Aishwarya Jayagopal, Sagar Gubbi, Subhashini Venugopalan, Alok Talekar, Vaibhav Rajan

Download BibTeX Cite

Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence. AI4Tech: AI Enabling Technologies. Pages 6546-6554.

https://doi.org/10.24963/ijcai.2026/728

Keywords

Domain-specific AI4Tech: AI4Agriculture

Abstract

Agricultural landscape segmentation in the Global South is challenging as it is characterized by fragmented plots, high intra-class variance, and a scarcity of labeled training data. Recent advances in segmentation have been made by Multimodal Large Language Models (MLLMs). However, current approaches encounter critical context length bottlenecks and a domain alignment gap in understanding satellite features. We address these limitations through MAgSeg, a novel, decoder-free MLLM segmentation approach. MAgSeg is an architecturally efficient approach that enables standard MLLMs to perform segmentation of complex smallholder agricultural landscapes from high-resolution satellite imagery, without requiring auxiliary vision decoders. We introduce a novel instruction tuning data format designed to enable scalable fine-tuning and post-training on high resolution satellite imagery, which enables MAgSeg to learn from the global context of the image while generating text tokens for only a patch within the image. Extensive evaluations on datasets spanning three countries in the Global South demonstrate that MAgSeg significantly outperforms state-of-the-art MLLM baselines, offering a scalable solution to map smallholder agricultural environments.

Piyush Tiwary, Utkarsh Ahuja, Depanshu Sani, Aishwarya Jayagopal, Sagar Gubbi, Subhashini Venugopalan, Alok Talekar, and Vaibhav Rajan. 2026. MAgSeg: Segmentation of Agricultural Landscapes in High-Resolution Satellite Imagery using Multimodal Large Language Models. In Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-26. IJCAI, 6546–6554. https://doi.org/10.24963/ijcai.2026/728

@inproceedings{ijcai2026p728,
  title     = {MAgSeg: Segmentation of Agricultural Landscapes in High-Resolution Satellite Imagery using Multimodal Large Language Models},
  author    = {Tiwary, Piyush and Ahuja, Utkarsh and Sani, Depanshu and Jayagopal, Aishwarya and Gubbi, Sagar and Venugopalan, Subhashini and Talekar, Alok and Rajan, Vaibhav},
  booktitle = {Proceedings of the Thirty-Fifth International Joint Conference on
               Artificial Intelligence, {IJCAI-26}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Diego Calvanese},
  pages     = {6546--6554},
  year      = {2026},
  month     = {8},
  note      = {AI4Tech: AI Enabling Technologies},
  doi       = {10.24963/ijcai.2026/728},
  url       = {https://doi.org/10.24963/ijcai.2026/728},
}

TY  - CPAPER
TI  - MAgSeg: Segmentation of Agricultural Landscapes in High-Resolution Satellite Imagery using Multimodal Large Language Models
AU  - Tiwary, Piyush
AU  - Ahuja, Utkarsh
AU  - Sani, Depanshu
AU  - Jayagopal, Aishwarya
AU  - Gubbi, Sagar
AU  - Venugopalan, Subhashini
AU  - Talekar, Alok
AU  - Rajan, Vaibhav
T2  - Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-26
PB  - International Joint Conferences on Artificial Intelligence Organization
PY  - 2026
SP  - 6546
EP  - 6554
DO  - 10.24963/ijcai.2026/728
UR  - https://doi.org/10.24963/ijcai.2026/728
ER  -