Deep semi-supervised approach based on consistency regularization and similarity learning for weeds classification

TL;DR

This work tackles weed species classification under limited labeled data by proposing a deep semi-supervised method that fuses consistency regularization with similarity learning through a ConvNeXt-based auto-encoder. The model leverages a Dense encoder–decoder with extensive skip-connections and optimizes a three-term loss that includes supervised, reconstruction-based consistency, and similarity objectives. Experiments on the DeepWeeds dataset show that the proposed approach outperforms fully supervised baselines, particularly when labeled data are scarce, and retains robustness under noisy inference conditions. The results demonstrate practical impact for precision agriculture, enabling robust weed recognition with reduced labeling effort and improved resilience to real-world field variations.

Abstract

Weed species classification represents an important step for the development of automated targeting systems that allow the adoption of precision agriculture practices. To reduce costs and yield losses caused by their presence. The identification of weeds is a challenging problem due to their shared similarities with crop plants and the variability related to the differences in terms of their types. Along with the variations in relation to changes in field conditions. Moreover, to fully benefit from deep learning-based methods, large fully annotated datasets are needed. This requires time intensive and laborious process for data labeling, which represents a limitation in agricultural applications. Hence, for the aim of improving the utilization of the unlabeled data, regarding conditions of scarcity in terms of the labeled data available during the learning phase and provide robust and high classification performance. We propose a deep semi-supervised approach, that combines consistency regularization with similarity learning. Through our developed deep auto-encoder architecture, experiments realized on the DeepWeeds dataset and inference in noisy conditions demonstrated the effectiveness and robustness of our method in comparison to state-of-the-art fully supervised deep learning models. Furthermore, we carried out ablation studies for an extended analysis of our proposed joint learning strategy.

Figures (10)

• Figure 1: Overview of the semi-supervised approach
• Figure 2: Resulting images reconstructed from noisy inputs through our proposed deep encoder-decoder model with similarity learning
• Figure 3: Average performance of the deep learning models (semi-supervised and supervised) on the test subset, based on the Accuracy metric, with different sizes of the labeled training subset. For 20% labeled training subset, the results of the supervised models and ConvNeXt-Base-SSL were reported from bib70
• Figure 4: Average performance of the deep learning models (semi-supervised and supervised) on the test subset, based on the F1-Score metric, with different sizes of the labeled training subset
• Figure 5: Average performance(Accuracy and F1-Score) of the deep learning models on the test subset, with added gaussian noise to the inputs. Data partitioning: 15% labeled data, 20% validation, 20% test