Deep Learning for Precision Agriculture: Post-Spraying Evaluation and Deposition Estimation

Harry Rogers,Tahmina Zebin,Grzegorz Cielniak,Beatriz De La Iglesia,Ben Magri

2024-09-25

Abstract:Precision spraying evaluation requires automation primarily in post-spraying imagery. In this paper we propose an eXplainable Artificial Intelligence (XAI) computer vision pipeline to evaluate a precision spraying system post-spraying without the need for traditional agricultural methods. The developed system can semantically segment potential targets such as lettuce, chickweed, and meadowgrass and correctly identify if targets have been sprayed. Furthermore, this pipeline evaluates using a domain-specific Weakly Supervised Deposition Estimation task, allowing for class-specific quantification of spray deposit weights in {\mu}L. Estimation of coverage rates of spray deposition in a class-wise manner allows for further understanding of effectiveness of precision spraying systems. Our study evaluates different Class Activation Mapping techniques, namely AblationCAM and ScoreCAM, to determine which is more effective and interpretable for these tasks. In the pipeline, inference-only feature fusion is used to allow for further interpretability and to enable the automation of precision spraying evaluation post-spray. Our findings indicate that a Fully Convolutional Network with an EfficientNet-B0 backbone and inference-only feature fusion achieves an average absolute difference in deposition values of 156.8 {\mu}L across three classes in our test set. The dataset curated in this paper is publicly available at <a class="link-external link-https" href="https://github.com/Harry-Rogers/PSIE" rel="external noopener nofollow">this https URL</a>

Computer Vision and Pattern Recognition,Machine Learning

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **Automate post - spraying evaluation in precision agriculture and generate deposition amount estimates for each type of target without using traditional agricultural methods (such as water - sensitive paper and tracers)**.

Specifically, current evaluation methods for precision spraying systems mainly rely on human intervention, which not only increases costs but may also lead to subjective errors and decision - making delays. Traditional evaluation methods include:

1. **Water - Sensitive Papers (WSPs)**: They require manual placement and retrieval for analysis, and their texture is different from that of the actual target, resulting in potentially inaccurate estimated spraying deposition amounts.
2. **Tracers**: Sprays are marked with dyes, but different chemical applications may require different tracers, and in some cases, the target crops need to be harvested for analysis.

In addition, evaluation methods for some specific spraying systems have limitations, such as manually counting the number of sprayed weeds or simply assuming hitting the target, etc. These methods lack universality and accuracy.

To solve these problems, the author proposes a computer vision pipeline based on deep learning and Explainable Artificial Intelligence (XAI) for automatic post - spraying evaluation and deposition amount estimation. This pipeline can:

- **Semantically segment potential targets** (such as lettuce, brome grass, and Kentucky bluegrass) and correctly identify whether they have been sprayed.
- **Quantify the spraying deposition weight (in microliters, µL) in a class - specific manner through weakly - supervised deposition amount estimation tasks**.
- **Compare different Class Activation Mapping (CAM) techniques** (such as AblationCAM and ScoreCAM) to determine which method is more effective and more interpretable.
- **Introduce inference - dedicated feature fusion, combine auxiliary outputs with traditional outputs, and improve the interpretability and segmentation performance of the model**.

Ultimately, this research aims to develop a fully - automated process that can generate class - specific deposition amount estimates without traditional methods after spraying, thereby improving the evaluation efficiency and accuracy of precision spraying systems.