A Many-to-Many Fully Convolutional Recurrent Network for Multitemporal Crop Recognition

Abstract

<p><strong>Abstract.</strong> Recently, recurrent neural networks have been proposed for crop mapping from multitemporal remote sensing data. Most of these proposals have been designed and tested in temperate regions, where a single harvest per season is the rule. In tropical regions, the favorable climate and local agricultural practices, such as crop rotation, result in more complex spatio-temporal dynamics, where the single harvest per season assumption does not hold. In this context, a demand arises for methods capable of recognizing agricultural crops at multiple dates along the multitemporal sequence. In the present work, we propose to adapt two recurrent neural networks, originally conceived for single harvest per season, for multidate crop recognition. In addition, we propose a novel multidate approach based on bidirectional fully convolutional recurrent neural networks. These three architectures were evaluated on public Sentinel-1 data sets from two tropical regions in Brazil. In our experiments, all methods achieved state-of-the-art accuracies with a clear superiority of the proposed architecture. It outperformed its counterparts in up to 3.8% and 7.4%, in terms of per-month overall accuracy, and it was the best performing method in terms of F1-score for most crops and dates on both regions.</p>

Enhanced Convolutional-Neural-Network Architecture for Crop Classification

“…Two sequences of the Campo Verde database are considered: 1. from October 2015 to February 2016; and 2. from March to July 2016. The proposed methodology achieves a f 1 score of 75.89%, which is higher than the previous results reported in the literature [20,[30][31][32][33][34]. Indeed, a competitive rise of the classification rate in contrast to benchmark-and-recent works conducting experiments on the same dataset was accomplished.…”

“…As a result, it was observed that the accuracy was better when using DL techniques (specifically, CNN), rather than the other considered methods-indeed, the CNN-based approach showed a more stable behavior. The research presented in [33,34] apply DL techniques to classify not only Campo Verde but also Luis Eduardo database [45] (another agriculture database from Brazil). Both works use approaches based on an FCN and similar parameters including a 32 × 32 patch size and assess the methodology over individual images and sequences composed by several images.…”

“…Aimed at establishing a way forward to provide solutions in this regard, this work introduces a competitive methodology for crop classification from multispectral satellite imagery by taking advantage of using an enhanced 2D-CNN together with a novel postprocessing step. Inspired by the main workflow of various RNN/2D-CNN-based methods for patch classification [20,[30][31][32][33][34], the proposed methodology mainly contains four steps: image stacking, patch extraction, 2D-CNN based modelling for classification purposes and post-processing, as depicted in Figure 1. In broad terms, the proposed methodology works as follows: (1) the images are stacked to increase the number of features; (2) the input images are split into patches and fed into the 2D-CNN; (3) the 2D-CNN architecture is designed following a smaller-scale approach, and trained to recognize 10 different types of crops (soybean, corn, cotton, sorghum, non-commercial crops (NCC), pasture, eucalyptus, grass, soil, and Cerrado); and (4) reduce the classification error caused by lower-spatial-resolution images, and a here-introduced post-processing step is performed.…”

First Results of the Lem Benchmark Database for Agricultural Applications

“…The data available on the LEM benchmark database is being used in the development of PhD thesis, master dissertations, and other studies, resulting in six publications so far (Sanches et al, 2018a;Achanccaray, 2019;Chamorro, 2019;Chamorro et al, 2019, Dutra et al, 2019Prudente et al, 2019), with interesting results. And future studies are being planned to collect more data from LEM municipality regarding another Brazilian crop year and other satellites (commercial radar data) to improve the LEM database.…”

“…3.2.2 Many-to-many fully convolutional recurrent networks for multitemporal crop recognition using SAR image sequences: As stated before, tropical regions, as in the LEM database, present more complex spatio-temporal dynamics than temperate regions, where there is a single harvest per season. In this sense, Chamorro et al (2019) adapted two recurrent neural networks (RNNs), originally conceived for single harvest per season, for multidate crop recognition. In addition, a novel multidate approach based on bidirectional fully convolutional recurrent neural networks was proposed.…”