Crop disease detection is important due to its significant impact on agricultural productivity and global food security. Traditional disease detection methods often rely on labour‐intensive field surveys and manual inspection, which are time‐consuming and prone to human error. In recent years, the advent of imaging technologies coupled with machine learning (ML) algorithms has offered a promising solution to this problem, enabling rapid and accurate identification of crop diseases. Previous studies have demonstrated the potential of image‐based techniques in detecting various crop diseases, showcasing their ability to capture subtle visual cues indicative of pathogen infection or physiological stress. However, the field is rapidly evolving, with advancements in sensor technology, data analytics and artificial intelligence (AI) algorithms continually expanding the capabilities of these systems. This review paper consolidates the existing literature on image‐based crop disease detection using ML, providing a comprehensive overview of cutting‐edge techniques and methodologies. Synthesizing findings from diverse studies offers insights into the effectiveness of different imaging platforms, contextual data integration and the applicability of ML algorithms across various crop types and environmental conditions. The importance of this review lies in its ability to bridge the gap between research and practice, offering valuable guidance to researchers and agricultural practitioners.
Plant disease outbreaks continuously challenge food security and sustainability. Traditional chemical methods used to treat diseases have environmental and health concerns, raising the need to enhance inherent plant disease resistance mechanisms. Traits, including disease resistance, can be linked to specific loci in the genome and identifying these markers facilitates targeted breeding approaches. Several methods, including genome‐wide association studies and genomic selection, have been used to identify important markers and select varieties with desirable traits. However, these traditional approaches may not fully capture the non‐linear characteristics of the effect of genomic variation on traits. Machine learning, known for its data‐mining abilities, offers an opportunity to enhance the accuracy of the existing trait association approaches. It has found applications in predicting various agronomic traits across several species. However, its use in disease resistance prediction remains limited. This review highlights the potential of machine learning as a complementary tool for predicting the genetic loci contributing to pathogen resistance. We provide an overview of traditional trait prediction methods, summarize machine‐learning applications, and address the challenges and opportunities associated with machine learning‐based crop disease resistance prediction.
Weeds pose a significant threat to agricultural production, leading to substantial yield losses and increased herbicide usage, with severe economic and environmental implications. This paper uses deep learning to explore a novel approach via targeted segmentation mapping of crop plants rather than weeds, focusing on canola (Brassica napus) as the target crop. Multiple deep learning architectures (ResNet-18, ResNet-34, and VGG-16) were trained for the pixel-wise segmentation of canola plants in the presence of other plant species, assuming all non-canola plants are weeds. Three distinct datasets (T1_miling, T2_miling, and YC) containing 3799 images of canola plants in varying field conditions alongside other plant species were collected with handheld devices at 1.5 m. The top performing model, ResNet-34, achieved an average precision of 0.84, a recall of 0.87, a Jaccard index (IoU) of 0.77, and a Macro F1 score of 0.85, with some variations between datasets. This approach offers increased feature variety for model learning, making it applicable to the identification of a wide range of weed species growing among canola plants, without the need for separate weed datasets. Furthermore, it highlights the importance of accounting for the growth stage and positioning of plants in field conditions when developing weed detection models. The study contributes to the growing field of precision agriculture and offers a promising alternative strategy for weed detection in diverse field environments, with implications for the development of innovative weed control techniques.
The timing of flowering in soybean [ Glycine max (L.) Merr.], a key legume crop, is influenced by many factors, including daylight length or photoperiodic sensitivity, that affect crop yield, productivity, and geographical adaptation. Despite its importance, a comprehensive understanding of the local linkage landscape and allelic diversity within regions of the genome influencing flowering and contributing to phenotypic variation in subpopulations has been limited. This study addresses these gaps by conducting an in‐depth trait association and linkage analysis coupled with local haplotyping using advanced bioinformatics tools, including crosshap , to characterize genomic variation using a pangenome dataset representing 915 domesticated and wild‐type individuals. The association analysis identified eight significant loci on seven chromosomes. Moving beyond traditional association analysis, local haplotyping of targeted regions on chromosomes 6 and 20 identified distinct haplotype structures, variation patterns, and genomic candidates influencing flowering in subpopulations. These results suggest the action of a network of genomic candidates influencing flowering time and an untapped reservoir of genomic variation for this trait in wild germplasm. Notably, GlymaLee.20G147200 on chromosome 20 was identified as a candidate gene that may cause delayed flowering in soybean, potentially through histone modifications of floral repressor loci as seen in Arabidopsis thaliana (L.) Heynh. These findings support future functional validation of haplotype‐based alleles for marker‐assisted breeding and genomic selection to enhance latitude adaptability of soybean without compromising yield.
Gene models are regions of the genome that can be transcribed into RNA and translated to proteins, or belong to a class of non-coding RNA genes. The prediction of gene models is a complex process that can be unreliable, leading to false positive annotations. To help support the calling of confident conserved gene models and minimize false positives arising during gene model prediction we have developed Truegene, a machine learning approach to classify potential low confidence gene models using 14 gene and 41 protein-based characteristics. Amino acid and nucleotide sequence-based features were calculated for conserved (high confidence) and non-conserved (low confidence) annotated genes from the published Pisum sativum Cameor genome. These features were used to train eXtreme Gradient Boost (XGBoost) classifier models to predict whether a gene model is likely to be real. The optimized models demonstrated a prediction accuracy ranging from 87% to 90% and an F-1 score of 0.91–0.94. We used SHapley Additive exPlanations (SHAP) and feature importance plots to identify the features that contribute to the model predictions, and we show that protein and gene-based features can be used to build accurate models for gene prediction that have applications in supporting future gene annotation processes.