Plant Recognition Research Articles

Medical Plant Recognition and Classification using CNN

Medical Plant Recognition and Classification using CNN

KOALA: A Modular Dual-Arm Robot for Automated Precision Pruning Equipped with Cross-Functionality Sensor Fusion

Landscape maintenance is essential for ensuring agricultural productivity, promoting sustainable land use, and preserving soil and ecosystem health. Pruning is a labor-intensive task among landscaping applications that often involves repetitive pruning operations. To address these limitations, this paper presents the development of a dual-arm holonomic robot (called the KOALA robot) for precision plant pruning. The robot utilizes a cross-functionality sensor fusion approach, combining light detection and ranging (LiDAR) sensor and depth camera data for plant recognition and isolating the data points that require pruning. The You Only Look Once v8 (YOLOv8) object detection model powers the plant detection algorithm, achieving a 98.5% pruning plant detection rate and a 95% pruning accuracy using camera, depth sensor, and LiDAR data. The fused data allows the robot to identify the target boxwood plants, assess the density of the pruning area, and optimize the pruning path. The robot operates at a pruning speed of 10–50 cm/s and has a maximum robot travel speed of 0.5 m/s, with the ability to perform up to 4 h of pruning. The robot’s base can lift 400 kg, ensuring stability and versatility for multiple applications. The findings demonstrate the robot’s potential to significantly enhance efficiency, reduce labor requirements, and improve landscape maintenance precision compared to those of traditional manual methods. This paves the way for further advancements in automating repetitive tasks within landscaping applications.

Medicinal Plant Recognition Based on the Seedling Image and Deep Learning

In the realm of botanical science and environmental conservation, identifying medicinal plant species accurately and efficiently is crucial for their sustainable cultivation and conservation. This paper introduces a novel approach that leverages seedling images and deep learning techniques to develop a computer vision system. The system aims to recognize medicinal plants at their early growth stages, a task that is fundamental for effective plant management and preservation efforts. The proposed work utilizes the Inception 3, a deep convolutional neural network (CNN) architecture, trained on a comprehensive dataset of seedling images from diverse medicinal plant species. Evaluating the system's performance with various metrics reveals its exceptional accuracy, reaching up to 97.6% in identifying medicinal plant species. This model holds promise for applications in plant conservation, biodiversity assessment, and the cultivation of medicinal plants.

Children’s moral judgments and reasoning regarding environmentally harmful behaviors: Variation by victim type and moderation effect of connectedness to nature

This study explored the differences in young children’s moral judgments regarding environmentally harmful behaviors based on victim type and the moderation effect of connectedness to nature (CN). Further, this study investigated which reasoning types (ecocentric, anthropocentric, and non-environmental) significantly explained children’s moral judgment. This study included 185 five-year-old Korean children and their mothers. The children participated in two interview sessions. In one session, they were randomly assigned to one of five groups—no video or video with unspecified, human, animal, or plant victim. In another session, they completed a Korean sentence comprehension test and a task assessing their recognition of animals and plants from the video. Mothers responded to a survey on their children’s CN and engagement in pro-environmental activities. A generalized Linear Mixed Model was employed using SPSS Win 27.0. Children’s judgment of environmentally harmful behaviors was more negative when presented with unspecified or human victims. The reasoning for their moral judgments and the moderating role of their CN varied by victim type. Environmental reasoning, especially ecocentric reasoning, was crucial in explaining children’s moral judgments of environmentally harmful behaviors. Thus, considering victims and environmental reasoning types is important for environmental education for sustainable development for young children, emphasizing the need to enhance children’s CN.

The reduction of chemical inputs by ultra-precise smart spot sprayer technology maximizes crop potential by lowering phytotoxicity

Weeding is at the heart of agriculture. Today, using herbicides is unavoidable in conventional agriculture. However, increasing ecological awareness, the need to reduce carbon dioxide emissions, and the pursuit of sustainable food have driven industries and academics to explore new approaches for growing and protecting crops. In this context, smart spot sprayers emerge as a highly innovative alternative to broadcast treatments, which overuse pesticides, and mechanical weeding, which is limited by restrictive application conditions. Using artificial intelligence (AI) for plant recognition and ultra-precise spraying technology, herbicides are applied directly to weeds with minimal overspray, reducing their use by 90% compared to conventional methods. Spot sprayers' narrowly targeted spray limits chemicals' contact with the crop, resulting in low phytotoxicity, preserving natural crop development while maximizing yield potential. In addition, site-specific application increases the selectivity of chemical products artificially, enabling more concentrated mixtures to be used. Also, AI enables the selective application of non-selective molecules, supporting their use in established cultures. These characteristics open new opportunities for effective weed control where the traditional solutions fall short. This publication demonstrates the potential of the ultra-precise smart spot sprayer ARA in weed management in onion and sugar beet crops. Although comparing the overall performance of ARA with other smart spot sprayers on the market cannot be done in this article, this technology's potential in rendering weed management on farms more effective, sustainable, and saving labor without compromising yield is nevertheless indicated.

Characterization of Mung Plants Treated with Iron Oxide Nanoparticles Using Raman and Ultraviolet-Visible Spectroscopy Coupled with Chemometrics

In order to meet the demands of the expanding population, nanotechnology-based materials are being investigated for a potential solution to enhance crop yield. This study examines the effect of different concentrations (0.2, 0.6, 1.0, and 1.4 mM) of iron oxide nanoparticles (Fe2O3 NPs) on mung plants using confocal micro Raman and ultraviolet-visible spectroscopy. The analysis shows that the biochemical profile of the plants is altered. The increased intensity values of carotenoids, pectin, lignin, proteins, cellulose, carbohydrates, and aliphatics observed in Raman spectroscopy, along with increased concentrations of photosynthetic pigments in ultraviolet-visible spectroscopy, suggest that Fe2O3 NPs positively impact mung. In agro-related studies, Raman spectroscopy in assistance with chemometric analysis provides a reliable, noninvasive, fast, and label-free recognition and classification of plants under biotic or abiotic stress. Principal component analysis (PCA), linear discriminant analysis (LDA), and principal component analysis-linear discriminant analysis (PCA-LDA) were applied to the Raman spectra of treated and untreated mung leaves for rapid categorization and precise prediction. The results show that PCA-LDA achieved superior predicting performance and visualization, with an accuracy of 70%, along with higher recall, precision, f1 score, and area under receiver operating characteristics curves compared to LDA. The study reveals that spectroscopic techniques combined with chemometrics have great potential for confirmatory, nondestructive, and label-free assessments of biotic and abiotic stresses in plants.

Cowpea lipid transfer protein 1 regulates plant defense by inhibiting the cysteine protease of cowpea mosaic virus

Many virus genomes encode proteases that facilitate infection. The molecular mechanism of plant recognition of viral proteases is largely unexplored. Using the system of Vigna unguiculata and cowpea mosaic virus (CPMV), we identified a cowpea lipid transfer protein (LTP1) which interacts with CPMV-encoded 24KPro, a cysteine protease, but not with the enzymatically inactive mutant 24KPro(C166A). Biochemical assays showed that LTP1 inhibited 24KPro proteolytic cleavage of the coat protein precursor large coat protein-small coat protein. Transient overexpression of LTP1 in cowpea reduced CPMV infection, whereas RNA interference-mediated LTP1 silencing increased CPMV accumulation in cowpea. LTP1 is mainly localized in the apoplast of uninfected plant cells, and after CPMV infection, most of the LTP1 is relocated to intracellular compartments, including chloroplast. Moreover, in stable LTP1-transgenic Nicotiana benthamiana plants, LTP1 repressed soybean mosaic virus (SMV) nuclear inclusion a protease activity, and accumulation of SMV was significantly reduced. We propose that cowpea LTP1 suppresses CPMV and SMV accumulation by directly inhibiting viral cysteine protease activity.

Ss4368: Pathogen-Associated Molecular Pattern for Inducing Plant Cell Death and Resistance to Phytophthora capsici.

Plant recognition of pathogen-associated molecular patterns (PAMPs) is pivotal in triggering immune responses, highlighting their potential as inducers of plant immunity. However, the number of PAMPs identified and applied in such contexts remains limited. In this study, we characterize a novel PAMP, designated Ss4368, which is derived from Scleromitrula shiraiana. Ss4368 is specifically distributed among a few fungal genera, including Botrytis, Monilinia, and Botryotinia. The transient expression of Ss4368 elicits cell death in a range of plant species. The signaling peptides, three conserved motifs, and cysteine residues (C46, C88, C112, C130, and C148) within Ss4368 are crucial for inducing robust cell death. Additionally, these signaling peptides are essential for the protein's localization to the apoplast. The cell death induced by Ss4368 and its homologous protein, Bc4368, is independent of the SUPPRESSOR OF BIR1-1 (SOBIR1), BRI1-ASSOCIATED KINASE-1 (BAK1), and salicylic acid (SA) pathways. Furthermore, the immune responses triggered by Ss4368 and Bc4368 significantly enhance the resistance of Nicotiana benthamiana to Phytophthora capsici. Therefore, we propose that Ss4368, as a novel PAMP, holds the potential for developing strategies to enhance plant resistance against P. capsici.

Indigenous-led designation and management of culturally significant species.

Indigenous peoples globally are actively seeking better recognition of plants and animals that are of cultural significance, which encompass both species and ecological communities. Acknowledgement and collaborative management of culturally significant entities in biodiversity conservation improves environmental outcomes as well as the health and wellbeing of Indigenous people. The global diversity and complexity of Indigenous knowledge, values and obligations make achieving a universal approach to designating culturally significant entities highly unlikely. Instead, empowering local Indigenous-led governance structures with methods to identify place-based culturally significant entities will yield culturally supported results. Here we used a structured decision-making framework with objectives and biocultural measures developed by Indigenous experts, with the aim of prioritizing place-based culturally significant entities for collaborative management approaches on Bundjalung Country in coastal eastern Australia. We found some congruence and some important differences between culturally significant entities priorities and management compared with the colonial focus of threatened species management underpinned by current laws and policies. We provide reproduceable methods and a demonstration of successful local culturally significant entities designation and prioritization in an Australian context that highlights opportunities for Indigenous leadership, supported by governments in the designation and management of culturally significant entities.

Viral Recognition and Evasion in Plants.

Viruses, causal agents of devastating diseases in plants, are obligate intracellular pathogens composed of a nucleic acid genome and a limited number of viral proteins. The diversity of plant viruses, their diminutive molecular nature, and their symplastic localization pose challenges to understanding the interplay between these pathogens and their hosts in the currently accepted framework of plant innate immunity. It is clear, nevertheless, that plants can recognize the presence of a virus and activate antiviral immune responses, although our knowledge of the breadth of invasion signals and the underpinning sensing events is far from complete. Below, I discuss some of the demonstrated or hypothesized mechanisms enabling viral recognition in plants, the step preceding the onset of antiviral immunity, as well as the strategies viruses have evolved to evade or suppress their detection.

Developing a smart forest protection mini program with Spring Boot: Integrating big data and image recognition technologies

Forests are essential for ecological stability yet pose significant management challenges due to their expansive and remote nature. Despite advancements in data acquisition, there is a lack of tools that effectively combine image recognition with big data analytics for real-time monitoring. This study introduces a "Smart Forest Protection" mini program, developed using Spring Boot, which integrates these technologies to facilitate immediate and efficient forest management. The program features intelligent plant recognition, user-driven incident reporting, waste recycling point displays, and a platform for disseminating forest conservation knowledge. Additionally, a big data visualization platform illustrates the spatiotemporal distribution of forests in Zhuhai, aiding governmental and environmental agencies in making informed decisions that enhance forest conservation and promote sustainable practices.

Application of deep neural network identification models based on garden environment perception in plant protection

Perception of garden environment plays a crucial role in the control of invasive plants. Timely and accurate detection of invasive plants can significantly enhance the effectiveness of their control measures. Neural network technology can effectively identify the presence of invasive plants in the environment so that effective measures can be taken to protect garden plants from invasive species. In this paper, we propose a deep bilinear transformed ResNeXt (DBT-ResNeXt) model that replaces ResNet with ResNeXt to construct a two-branch structure for invasive plant recognition. The model is trained and validated using a dataset constructed from publicly available images of invasive plants. Experimental results demonstrate that compared to various other models, DBT-ResNeXt exhibits superior performance, achieving a recognition accuracy of 94.66%.

An Improved Ningxia Desert Herbaceous Plant Classification Algorithm Based on YOLOv8.

Wild desert grasslands are characterized by diverse habitats, uneven plant distribution, similarities among plant class, and the presence of plant shadows. However, the existing models for detecting plant species in desert grasslands exhibit low precision, require a large number of parameters, and incur high computational cost, rendering them unsuitable for deployment in plant recognition scenarios within these environments. To address these challenges, this paper proposes a lightweight and fast plant species detection system, termed YOLOv8s-KDT, tailored for complex desert grassland environments. Firstly, the model introduces a dynamic convolutional KernelWarehouse method to reduce the dimensionality of convolutional kernels and increase their number, thus achieving a better balance between parameter efficiency and representation ability. Secondly, the model incorporates triplet attention into its feature extraction network, effectively capturing the relationship between channel and spatial position and enhancing the model's feature extraction capabilities. Finally, the introduction of a dynamic detection head tackles the issue related to target detection head and attention non-uniformity, thus improving the representation of the target detection head while reducing computational cost. The experimental results demonstrate that the upgraded YOLOv8s-KDT model can rapidly and effectively identify desert grassland plants. Compared to the original model, FLOPs decreased by 50.8%, accuracy improved by 4.5%, and mAP increased by 5.6%. Currently, the YOLOv8s-KDT model is deployed in the mobile plant identification APP of Ningxia desert grassland and the fixed-point ecological information observation platform. It facilitates the investigation of desert grassland vegetation distribution across the entire Ningxia region as well as long-term observation and tracking of plant ecological information in specific areas, such as Dashuikeng, Huangji Field, and Hongsibu in Ningxia.

Deep learning-based aquatic plant recognition technique and natural ecological aesthetics conservation

Aquatic plants play a crucial role in the construction and maintenance of ecological systems in landscaping, contributing significantly to the enhancement and preservation of garden ecosystems. Monitoring the growth and population structure of aquatic plants enables the assessment of nutrient enrichment and water quality pollution in water bodies, serving as a basis for water quality improvement and protection. However, the lack of efficient automated monitoring technologies limits the efficiency of monitoring and fails to meet the demands of extensive and long-term monitoring. Deep learning techniques offer a promising solution for processing and analyzing aquatic plant monitoring data. By training deep learning models, automation and analysis of aquatic plant monitoring data can be achieved, thereby enhancing data processing efficiency and accuracy. In this study, aquatic plant image data were obtained through the internet to construct a dataset for aquatic plant recognition. We improved the DenseNet169 model to construct the EFL-DenseNet model. After initial training on the ImageNet dataset, transfer learning was applied to adapt the model to the self-built aquatic plant recognition dataset. The final model achieved an accuracy of 91.52%, demonstrating significant advantages over other models.

How do young identify plants? Using the drawing method to explore early ethnobotanical knowledge in Madagascar

Abstract In small‐scale societies, people learn to identify plant species during childhood. Plant recognition is an important baseline knowledge, immediately useful to avoid intoxication risk due to wrong identification. Plant recognition is the basis of other ethnobotanical knowledge essential for safeguarding biocultural diversity. However, despite many studies on folk classification, we still have a narrow understanding of the criteria locally used for species identification; the gap being even larger regarding children's plant identification criteria. Here, we study the criteria used by Betsileo children and adolescents to identify wild edible plant (WEP) species using a child‐adapted method including drawings and follow‐up interviews. We worked with 80 teenagers (from 12 to 17 years old; 51 girls, and 29 boys). Our results suggest that teenagers use a large spectrum of visual criteria to identify plants and that these criteria match with botanical and ecological knowledge documented in the literature and herbarium vouchers. We found that 35% of the identification criteria used were non‐morphological (e.g. phenology, biotic interactions), suggesting deep ecological knowledge. On average, teenagers use more than nine distinct criteria per plant, which allows them to identify most plant species with a very high level of precision. The precision level of plant representation increases with age for boys, but remains constant for girls, suggesting different dynamics in plant identification knowledge acquisition. We also found that boys and girls use different identification criteria: girls focus on morphological criteria while boys also incorporate ecological criteria, such as landscape features and biotic interactions, in their spectrum of identification keys. Our results highlight the complexity of teenagers' plant knowledge and the importance of the ecological context and gender in plant identification's knowledge acquisition. This knowledge acquired very early in childhood, constitutes the foundation of future interactions with nature and should be at the heart of environmental humanities studies and knowledge co‐production projects to tackle socio‐ecological concerns. Hence, we urge further research to explore innovative methods that complement traditional ethnoecological tools and capture complex sensory aspects of folk children's taxonomy to better understand human‐plant interactions and knowledge. Read the free Plain Language Summary for this article on the Journal blog.

Exploring the binding affinity and characteristics of DcitOBP9 in citrus psyllids

Odorant-binding proteins (OBPs) are crucial in insect olfaction. The most abundant expressed OBP of citrus psyllids, DcitOBP9 encodes 148 amino acids. DcitOBP9 lacks a transmembrane structure and possesses a 17-amino acid signal peptide at the N-terminus. Characterized by the six conserved cysteine sites, DcitOBP9 is classified as the Classical-OBP family. RT-qPCR experiments revealed ubiquitous expression of DcitOBP9 across all developmental stages of the citrus psyllid, with predominant expression in adults antennae. Fluorescence competitive binding assays demonstrated DcitOBP9′s strong affinity for ocimene, linalool, dodecanoic acid, and citral, and moderate affinity for dimethyl trisulfide. Additionally, it binds to myrcia, (−)-trans-caryophyllene, (±)-Citronellal, nonanal, and (+)-α-pinene. Among them, ocimene, linalool, and dodecanoic acid were dynamically bound to DcitOBP9, while citral was statically bound to DcitOBP9. Molecular docking simulations with the top five ligands indicated that amino acid residues V92, S72, P128, L91, L75, and A76 are pivotal in the interaction between DcitOBP9 and these odorants. These findings suggest DcitOBP9′s involvement in the citrus psyllid's host plant recognition and selection behaviors, thereby laying a foundation for elucidating the potential physiological and biological functions of DcitOBP9 and developing attractants.

Learning a complex network representation for shape classification

Shape contour is a key low-level characteristic, making shape description an important aspect in many computer vision problems, with several challenges such as variations in scale, rotation, and noise. In this paper, we introduce an approach for shape analysis and classification from binary images based on representations learned by applying Randomized Neural Networks (RNNs) on feature maps derived from a Complex Network (CN) framework. Our approach models the contour in a complex network and computes their topological measures using a dynamic evolution strategy. This evolution of the CN provides significant information into the physical aspects of the shape’s contour. Therefore, we propose embedding the topological measures computed from the dynamics of the CN evolution into a matrix representation, which we have named the Topological Feature Map (TFM). Then, we employ the RNN to learn representations from the TFM through a sliding window strategy. The proposed representation is formed by the learned weights between the hidden and output layers of the RNN. Our experimental results show performance improvements in shape classification using the proposed method across two generic shape datasets. We also applied our approach to the recognition of plant leaves, achieving high performance in this challenging task. Furthermore, the proposed approach has demonstrated robustness to noise and invariance to transformations in scale and orientation of the shapes.

Protein degrons and degradation: Exploring substrate recognition and pathway selection in plants.

Proteome composition is dynamic and influenced by many internal and external cues, including developmental signals, light availability, or environmental stresses. Protein degradation, in synergy with protein biosynthesis, allows cells to respond to various stimuli and adapt by reshaping the proteome. Protein degradation mediates the final and irreversible disassembly of proteins, which is important for protein quality control and to eliminate misfolded or damaged proteins, as well as entire organelles. Consequently, it contributes to cell resilience by buffering against protein or organellar damage caused by stresses. Moreover, protein degradation plays important roles in cell signaling, as well as transcriptional and translational events. The intricate task of recognizing specific proteins for degradation is achieved by specialized systems that are tailored to the substrate's physicochemical properties and subcellular localization. These systems recognize diverse substrate cues collectively referred to as "degrons," which can assume a range of configurations. They are molecular surfaces recognized by E3 ligases of the ubiquitin-proteasome system but can also be considered as general features recognized by other degradation systems, including autophagy or even organellar proteases. Here we provide an overview of the newest developments in the field, delving into the intricate processes of protein recognition and elucidating the pathways through which they are recruited for degradation.

Interactive transcriptome analyses of Northern Wild Rice (Zizania palustris L.) and Bipolaris oryzae show convoluted communications during the early stages of fungal brown spot development.

Fungal diseases, caused mainly by Bipolaris spp., are past and current threats to Northern Wild Rice (NWR) grain production and germplasm preservation in both natural and cultivated settings. Genetic resistance against the pathogen is scarce. Toward expanding our understanding of the global gene communications of NWR and Bipolaris oryzae interaction, we designed an RNA sequencing study encompassing the first 12 h and 48 h of their encounter. NWR activated numerous plant recognition receptors after pathogen infection, followed by active transcriptional reprogramming of signaling mechanisms driven by Ca2+ and its sensors, mitogen-activated protein kinase cascades, activation of an oxidative burst, and phytohormone signaling-bound mechanisms. Several transcription factors associated with plant defense were found to be expressed. Importantly, evidence of diterpenoid phytoalexins, especially phytocassane biosynthesis, among expression of other defense genes was found. In B. oryzae, predicted genes associated with pathogenicity including secreted effectors that could target plant defense mechanisms were expressed. This study uncovered the early molecular communication between the NWR-B. oryzae pathosystem, which could guide selection for allele-specific genes to boost NWR defenses, and overall aid in the development of more efficient selection methods in NWR breeding through the use of the most virulent fungal isolates.

Adapting the Segment Anything Model for Plant Recognition and Automated Phenotypic Parameter Measurement

Traditional phenotyping relies on experts visually examining plants for physical traits like size, color, or disease presence. Measurements are taken manually using rulers, scales, or color charts, with all data recorded by hand. This labor-intensive and time-consuming process poses a significant obstacle to the efficient breeding of new cultivars. Recent innovations in computer vision and machine learning offer potential solutions for accelerating the development of robust and highly effective plant phenotyping. This study introduces an efficient plant recognition framework that leverages the power of the Segment Anything Model (SAM) guided by Explainable Contrastive Language–Image Pretraining (ECLIP). This approach can be applied to a variety of plant types, eliminating the need for labor-intensive manual phenotyping. To enhance the accuracy of plant phenotype measurements, a B-spline curve is incorporated during the plant component skeleton extraction process. The effectiveness of our approach is demonstrated through experimental results, which show that the proposed framework achieves a mean absolute error (MAE) of less than 0.05 for the majority of test samples. Remarkably, this performance is achieved without the need for model training or labeled data, highlighting the practicality and efficiency of the framework.