Harnessing Time-Series Satellite Data for Crop Classification: A Multi-Model Approach

DOI

10.2991/978-94-6463-858-5_10How to use a DOI?

Keywords

Self supervised; Transformers; BERT (Bi-directional Encoder Representations from Transformers); CNN; RNN; LSTM; SVM

Abstract

The satellite image time series classification plays a crucial role in remote sensing and offers multiple applications for agriculture, land management, and disaster monitoring. Although deep learning methods are powerful, they suffer from overfitting in some cases, especially when there is a lack of labeled data. In this paper, we introduce a pipeline that includes the pre-processing step followed by BERT-based self-supervised pre-training and then fine-tuning. Our methodology starts with converting the .npz files to the .csv format to make it compatible with the training procedure. We pretrain a transformer-based model on massive amounts of un labelled data, identifying corrupted observations within the time series. We then fine-tune this pretrained model on designated tasks and thereby achieve accuracy and generalization. We compare our approach with regular models, including CNNs, RNNs, LSTMs, and SVMs, both with and without pretraining. Our experiments using California Labelled datasets demonstrate the effectiveness of the approach we have proposed.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - T. Lakshmi Praveena
AU  - Kumar Rajamani
AU  - Chelimela Roshini
AU  - Nallamothu Himasri
AU  - Erasani Chitra
PY  - 2025
DA  - 2025/11/04
TI  - Harnessing Time-Series Satellite Data for Crop Classification: A Multi-Model Approach
BT  - Proceedings of International Conference on Computer Science and Communication Engineering (ICCSCE 2025)
PB  - Atlantis Press
SP  - 101
EP  - 107
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-858-5_10
DO  - 10.2991/978-94-6463-858-5_10
ID  - Praveena2025
ER  -