Multi-modal Contrastive Learning for Crop Classification Using Sentinel2 and Planetscope
Bibliographic record
Abstract
Remote sensing has enabled large-scale crop classification to understand agricultural ecosystems and estimate production yields. Since a few years, machine learning is increasingly used for automated crop classification. However, in most approaches the novel algorithms are applied to custom datasets containing information of few crop fields covering a small region and this often leads to models that lack generalization capability. In this work, we propose a multi-modal contrastive self-supervised learning approach to obtain a pre-trained model for crop-classification without the use of labeled data. Such multi-modal self-supervised learning exploits the synergies of different data sources to obtain a richer representation of the data. We build our analysis by adapting the DENETHOR dataset developed for a part of Eastern Germany to our usecase. We use the publicly available Sentinel2 and commercial Planetscope data. While Sentinel2 has higher spectral resolution, Planetscope has finer spatial resolution. For an end-user application, only one source is required. In this work, we analyze and compare the performance of our multi-modal self-supervised model against the uni-modal contrastive self-supervised model using the SCARF algorithm. In addition, we also compare our multi-modal self-supervised model with a supervised model. We find that our multi-modal pre-trained model surpasses the uni-modal and supervised models in almost all test cases.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".