Context-self contrastive pretraining for crop type semantic segmentation

TL;DR

This paper introduces a contrastive pre-training method called CSCL for crop type segmentation in satellite images, significantly improving boundary accuracy and providing a large annotated dataset for the community.

Contribution

The paper presents a novel contrastive learning scheme tailored for dense classification and releases the largest annotated SITS dataset for crop segmentation.

Findings

CSCL improves boundary segmentation performance.

The dataset enables more granular crop classification.

Pre-training with CSCL enhances baseline models.

Abstract

In this paper, we propose a fully supervised pre-training scheme based on contrastive learning particularly tailored to dense classification tasks. The proposed Context-Self Contrastive Loss (CSCL) learns an embedding space that makes semantic boundaries pop-up by use of a similarity metric between every location in a training sample and its local context. For crop type semantic segmentation from Satellite Image Time Series (SITS) we find performance at parcel boundaries to be a critical bottleneck and explain how CSCL tackles the underlying cause of that problem, improving the state-of-the-art performance in this task. Additionally, using images from the Sentinel-2 (S2) satellite missions we compile the largest, to our knowledge, SITS dataset densely annotated by crop type and parcel identities, which we make publicly available together with the data generation pipeline. Using that data we find CSCL, even with minimal pre-training, to improve all respective baselines and present a process for semantic segmentation at super-resolution for obtaining crop classes at a more granular level. The code and instructions to download the data can be found in https://github.com/michaeltrs/DeepSatModels.