Context-self contrastive pretraining for crop type semantic segmentation

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.

arXiv comment15 pages, 17 figures

References in corpus (15)

• Adam: A Method for Stochastic Optimization
• FaceNet: A Unified Embedding for Face Recognition and Clustering
• A Simple Framework for Contrastive Learning of Visual Representations
• Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting
• Representation Learning with Contrastive Predictive Coding
• Deep learning in agriculture: A survey
• Improved Baselines with Momentum Contrastive Learning
• The Cityscapes Dataset for Semantic Urban Scene Understanding
• Lip Reading Sentences in the Wild
• Land Cover Classification via Multi-temporal Spatial Data by Recurrent Neural Networks
• Multi-Temporal Land Cover Classification with Sequential Recurrent Encoders
• Fully Connected Deep Structured Networks
• Non-local Neural Networks
• Distance transform regression for spatially-aware deep semantic segmentation
• Dense Contrastive Learning for Self-Supervised Visual Pre-Training

Cited by in corpus (2)

• Deep Learning for Satellite Image Time Series Analysis: A Review
• Semi-Supervised Semantic Segmentation with Cross Teacher Training