Semi-supervised deep learning for multi-crop segmentation using Sentinel-1 and Sentinel-2 data in Haryana, India

SARAVANAKUMAR R; RAJNI  JAIN; ANSHU  BHARADWAJ; VINAY KUMAR  SEHGAL; ANKUR  BISWAS; ALKA  ARORA

doi:10.56093/ijas.v96i7.173401

Authors

SARAVANAKUMAR R ICAR - Indian Agricultural Research Institute, New Delhi, India
RAJNI JAIN ICAR-National Institute of Agricultural Economics and Policy Research, New Delhi 110 012, India
ANSHU BHARADWAJ ICAR-Indian Agricultural Statistical Research Institute, New Delhi, India
VINAY KUMAR SEHGAL ICAR - Indian Agricultural Research Institute, New Delhi, India
ANKUR BISWAS ICAR-Indian Agricultural Statistical Research Institute, New Delhi, India
ALKA ARORA ICAR-Indian Agricultural Statistics Research Institute, New Delhi

https://doi.org/10.56093/ijas.v96i7.173401

Keywords:

Crop segmentation, Mustard, Pseudo-labeling, Remote sensing, Wheat

Abstract

Accurate crop classification from satellite imagery is essential for agricultural monitoring; however, mapping the crop type in data-scarce regions remains challenging due to limited ground truth, spectral similarities between crops, and strong temporal variability. Supervised deep learning approaches are further constrained by their dependence on large annotated datasets. In this study, a semi-supervised framework based on R3Net a residual-recurrent refinement network, was developed to leverage both optical and SAR signals for pixel-level segmentation of winter (rabi) season crops [wheat (Triticum aestivum L.), mustard (Brassica juncea L.), and other classes] using multi-temporal satellite imagery. The architecture integrates parallel temporal encoders with recurrent units to capture spatial-spectral features and phenological dynamics across key growth stages. A confidence-aware pseudo-labeling strategy was introduced to effectively utilize large volumes of unlabeled data while suppressing uncertain predictions. The framework was evaluated against three baselines: fully supervised training, standard pseudo-labeling, and the mean teacher method. The supervised model achieved overall accuracy of 65.21% and a mean IoU of 48.55%, whereas the semi-supervised approaches significantly improved segmentation performance. Among all methods, the proposed confidence-aware R3Net achieved the best performance, with 76.05% overall accuracy and a mean IoU of 59.03%, including notable improvements in mustard segmentation. Testing on an independent geographic region (IARI 2022–23) further demonstrated robust generalisation capability, producing segmentation maps consistent with field boundaries and crop phenology.

Downloads

Download data is not yet available.

References

Aksu G, Güzeller C O and Eser M T. 2019. The effect of the normalisation method used in different sample sizes on the success of Artificial Neural Network Model. International Journal of Assessment Tools in Education 6(2): 170–92. https://doi.org/10.21449/ijate.479404 DOI: https://doi.org/10.21449/ijate.479404

Ashish. 2024. Comparative insights into the changing patterns of cropping intensity in Haryana. International Journal of English Literature and Social Sciences 9(6): 94. https://doi. org/10.22161/ijels.96.18 DOI: https://doi.org/10.22161/ijels.96.18

Bagaria N and Surendra S J. 2025. Growth dynamics and sustainability of major crops in Haryana, India. Indian Journal of Extension Education 61(2): 80–84. https://doi.org/10.48165/ IJEE.2025.61215 DOI: https://doi.org/10.48165/IJEE.2025.61215

Chen H, Li Z, Wu J, Xiong W and Du C. 2023. SemiRoadExNet: A semi-supervised network for road extraction from remote sensing imagery via adversarial learning. ISPRS Journal of Photogrammetry and Remote Sensing 198: 169–83. https://doi.org/10.1016/j.isprsjprs.2023.03.012 DOI: https://doi.org/10.1016/j.isprsjprs.2023.03.012

Chen G, Shi B, Zhang Y, He Z and Zhang P. 2024a. CGSNet: Cross- consistency guiding semi-supervised semantic segmentation network for remote sensing of plateau lake. Journal of Network and Computer Applications 230: 103974. https://doi.org/10.1016/j.jnca.2024.103974 DOI: https://doi.org/10.1016/j.jnca.2024.103974

Chen J, Chen G, Zhang L, Huang M, Luo J, Ding M and Ge Y. 2024b. Category-sensitive semi-supervised semantic segmentation framework for land-use/land-cover mapping with optical remote sensing images. International Journal of Applied Earth Observation and Geoinformation 134: 104160. https://doi.org/10.1016/j.jag.2024.104160 DOI: https://doi.org/10.1016/j.jag.2024.104160

Chen J, Sun B, Lyu X, Chen G, Wang L, Fang B, Chang Y, Li Y and Zhang J. 2022. Semi-supervised semantic segmentation framework with pseudo supervisions for land-use/land-cover mapping in coastal areas. International Journal of Applied Earth Observation and Geoinformation 112: 102881. https://doi.org/10.1016/j.jag.2022.102881 DOI: https://doi.org/10.1016/j.jag.2022.102881

Cheng Z, Meng J and Wang Y. 2016. Improving spring maize yield estimation at field scale by assimilating time-series HJ-1 CCD data into the WOFOST model using a new method with fast algorithms. Remote Sensing 8(4): 303. https://doi.org/10.3390/ rs8040303 DOI: https://doi.org/10.3390/rs8040303

Everingham M, Van Gool L, Williams C K I, Winn J and Zisserman A. 2010. The PASCAL Visual Object Classes (VOC) challenge. International Journal of Computer Vision 88(2): 303–38. https://doi.org/10.1007/s11263-009-0275-4 DOI: https://doi.org/10.1007/s11263-009-0275-4

Feng Y, Wang L, Jin J and Wang X. 2025. A semi-supervised multi-source remote sensing image classification network based on adaptive pseudo-label generation. Applied Soft Computing 175: 113055. https://doi.org/10.1016/j.asoc.2025.113055 DOI: https://doi.org/10.1016/j.asoc.2025.113055

Feng Z, Huang G and Chi D. 2020. Classification of the complex agricultural planting structure with a semi-supervised extreme learning machine framework. Remote Sensing 12(22): 3708. https://doi.org/10.3390/rs12223708 DOI: https://doi.org/10.3390/rs12223708

Hochreiter S and Schmidhuber J. 1997. Long short-term memory. Neural Computation 9(8): 1735–80. https://doi.org/10.1162/neco.1997.9.8.1735 DOI: https://doi.org/10.1162/neco.1997.9.8.1735

Huang W, Shi Y, Xiong Z and Zhu X X. 2024. Decouple and weight semi-supervised semantic segmentation of remote sensing images. ISPRS Journal of Photogrammetry and Remote Sensing 212: 13–26. https://doi.org/10.1016/j.isprsjprs.2024.04.010 DOI: https://doi.org/10.1016/j.isprsjprs.2024.04.010

Jones M O, Figueiredo G, Howson S, Toro A, Rundquist S, Garner G, Della Nave F, Delgado G, Yi Z-F, Ahn P, Barrett S J, Bader M, Rollend D, Bendixen T, Albrecht J, Sogomo K, Musse Z Z and Shriver J. 2024. Monitoring and mapping a decade of regenerative agricultural practices across the contiguous United States. Land 13(12): 2246. https://doi.org/10.3390/ land13122246 DOI: https://doi.org/10.3390/land13122246

Kumar V, Jat H S, Sharma P C, Balwinder-Singh, Gathala M K, Malik R K, Kamboj B R, Yadav A K, Ladha J K, Raman A, Sharma D K and McDonald A. 2018. Can productivity and profitability be enhanced in intensively managed cereal systems while reducing the environmental footprint of production? Assessing sustainable intensification options in the breadbasket of India. Agriculture, Ecosystems and Environment 252: 132–47. https://doi.org/10.1016/j.agee.2017.10.006 DOI: https://doi.org/10.1016/j.agee.2017.10.006

Kussul N, Lavreniuk M, Skakun S and Shelestov A. 2017. Deep learning classification of land cover and crop types using remote sensing data. IEEE Geoscience and Remote Sensing Letters 14(5): 778–82. https://doi.org/10.1109/LGRS.2017.2681128 DOI: https://doi.org/10.1109/LGRS.2017.2681128

Kwak T and Kim Y. 2023. Semi-supervised land cover classification of remote sensing imagery using CycleGAN and EfficientNet. KSCE Journal of Civil Engineering 27(4): 1760–73. https://doi.org/10.1007/s12205-023-2285-0 DOI: https://doi.org/10.1007/s12205-023-2285-0

Li Y, Li Z, Wang Z, Wang K and Yu Q. 2024. Semi-supervised remote sensing image scene classification with prototype-based consistency. Chinese Journal of Aeronautics 37(2): 459–70. https://doi.org/10.1016/j.cja.2023.12.012 DOI: https://doi.org/10.1016/j.cja.2023.12.012

Li Y, Wu T, Ge Y, Xi S, Zhang T and Zhang W. 2023. Semi- supervised cooperative regression model for small sample estimation of citrus leaf nitrogen content with UAV images. International Journal of Remote Sensing 44(23): 7237–62. https://doi.org/10.1080/01431161.2023.2240027 DOI: https://doi.org/10.1080/01431161.2023.2240027

Liu B, Li B and Li S. 2025. Multi-scale fusion pixel and instance contrastive self-supervised learning for semantic segmentation of high-resolution Earth surface images. Advances in Space Research 75: 1854–70. https://doi.org/10.1016/j.asr.2024.10.026 DOI: https://doi.org/10.1016/j.asr.2024.10.026

Luo J, Fan J, Huang S, Wu S, Zhang F and Li X. 2025. Semi- supervised learning techniques for detection of dead pine trees with UAV imagery for pine wilt disease control. International Journal of Remote Sensing 46(2): 575–605. DOI: https://doi.org/10.1080/01431161.2024.2421945

Menezes G K, Astolfi G, Martins J A C, Tetila E C, Oliveira Junior A da S, Gonçalves D N, Marcato Junior J, Silva J A, Li J, Gonçalves W N and Pistori H. 2023. Pseudo-label semi-supervised learning for soybean monitoring. Smart Agricultural Technology 4: 100216. https://doi.org/10.1016/j. atech.2023.10021 DOI: https://doi.org/10.1016/j.atech.2023.100216

Peña J M, Gutiérrez P A, Hervás-Martínez C, Six J, Plant R E and López-Granados F. 2014. Object-based image classification of summer crops with machine learning methods. Remote Sensing 6(6): 5019–41. https://doi.org/10.3390/rs6065019 DOI: https://doi.org/10.3390/rs6065019

Ramteke P L and Kshirsagar U. 2023. The role of machine intelligence in agriculture: A case study. Research Trends in Artificial Intelligence: Internet of Things 1: 54. https://doi.or g/10.2174/9789815136449123010007 DOI: https://doi.org/10.2174/9789815136449123010007

Ronneberger O, Fischer P and Brox T. 2015. U-Net: Convolutional networks for biomedical image segmentation. (In) Medical Image Computing and Computer-Assisted Intervention - MICCAI 2015, 9351, pp. 234–41. DOI: https://doi.org/10.1007/978-3-319-24574-4_28

Navab N, Hornegger J, Wells W M and Frangi A F (Eds). Song H, Zhuang T, Li X, Ruan G, Schepers J, Wang D, Liu X, Tian Y, Zhu Y, Cao W and Cao Q. 2025. Enhancing winter wheat growth indicator prediction with multi-task learning and multi-source data. European Journal of Agronomy 168: 127629. https://doi.org/10.1016/j.eja.2025.127629 DOI: https://doi.org/10.1016/j.eja.2025.127629

Sokolova M and Lapalme G. 2009. A systematic analysis of performance measures for classification tasks. Information Processing and Management 45(4): 427–37. https://doi. org/10.1016/j.ipm.2009.03.002 DOI: https://doi.org/10.1016/j.ipm.2009.03.002

Tian Y, Zhang L, Sun J, Yin G and Dong Y. 2022. Consistency regularisation teacher-student semi-supervised learning method for target recognition in SAR images. The Visual Computer 38: 4179–92. https://doi.org/10.1007/s00371-021-02287-z DOI: https://doi.org/10.1007/s00371-021-02287-z

Wang M, Lu S, Zhu D, Lin J and Wang Z. 2018. A high-speed and low-complexity architecture for softmax function in deep learning. (In) 2018 IEEE Asia Pacific Conference on Circuits and Systems (APCCAS), pp. 223–26. IEEE. https:// doi.org/10.1109/APCCAS.2018.8605654 DOI: https://doi.org/10.1109/APCCAS.2018.8605654

Wang S, Huang X, Han W, Li J, Zhang X and Wang L. 2023. Lithological mapping of geological remote sensing via adversarial semi-supervised segmentation network. International Journal of Applied Earth Observation and Geoinformation 125: 103536. https://doi.org/10.1016/j. jag.2023.103536 DOI: https://doi.org/10.1016/j.jag.2023.103536

Zhang L, Yang Z, Zhou G, Lu C, Chen A, Ding Y, Wang Y, Li L and Cai W. 2023. MDMASNet: A dual-task interactive semi-supervised remote sensing image segmentation method. Signal Processing 212: 109152. https://doi.org/10.1016/j. sigpro.2023.109152 DOI: https://doi.org/10.1016/j.sigpro.2023.109152