Annotation Techniques Research Articles

UAV photogrammetry for red tide monitoring and geo-localization via onboard GNSS/IMU data

As one of the worst ecological disasters in the world’s oceans, red tide seriously threatens marine ecology. Red tide monitoring is essential for preserving the ocean ecosystem. However, traditional photogrammetric processing workflows, e.g. SfM (Structure from Motion) and MVS (Multi-view Stereo) based image orientation and dense matching, do not apply to offshore images due to the low texture of ocean color. The primary contribution of this study is a UAV (Unmanned Aerial Vehicle)-based photogrammetric solution for red tide monitoring and direct geo-localization. First, a direct geo-localization model has been created, which solely uses onboard sensor data for the 3D coordinate calculation of 2D image targets based on the collinear equation. Second, two UAVs are chosen as photogrammetric systems that can supply the necessary data for the direct geo-localization model, including high-resolution images, camera intrinsic and extrinsic parameters. Third, a human annotation technique has been utilized to identify red tide regions whose ground polygons can be further estimated using the direct geo-localization model, taking into account how difficult it is to detect red tides automatically. Finally, the accuracy of direct geo-localization and the viability of the proposed solution has been evaluated and verified using real UAV datasets. The experimental results show that the direct geo-localization precision is better than 20 m when only onboard sensor data is used. The suggested workflow is appropriate for monitoring red tides, which can provide critical information for managing marine ecosystems. The executable toolkit would be made publicly available at https://github.com/json87/PhotoDigitizer.

Methods for improving colorectal cancer annotation efficiency for artificial intelligence-observer training.

Missing occult cancer lesions accounts for the most diagnostic errors in retrospective radiology reviews as early cancer can be small or subtle, making the lesions difficult to detect. Second-observer is the most effective technique for reducing these events and can be economically implemented with the advent of artificial intelligence (AI). To achieve appropriate AI model training, a large annotated dataset is necessary to train the AI models. Our goal in this research is to compare two methods for decreasing the annotation time to establish ground truth: Skip-slice annotation and AI-initiated annotation. We developed a 2D U-Net as an AI second observer for detecting colorectal cancer (CRC) and an ensemble of 5 differently initiated 2D U-Net for ensemble technique. Each model was trained with 51 cases of annotated CRC computed tomography of the abdomen and pelvis, tested with 7 cases, and validated with 20 cases from The Cancer Imaging Archive cases. The sensitivity, false positives per case, and estimated Dice coefficient were obtained for each method of training. We compared the two methods of annotations and the time reduction associated with the technique. The time differences were tested using Friedman's two-way analysis of variance. Sparse annotation significantly reduces the time for annotation particularly skipping 2 slices at a time (P < 0.001). Reduction of up to 2/3 of the annotation does not reduce AI model sensitivity or false positives per case. Although initializing human annotation with AI reduces the annotation time, the reduction is minimal, even when using an ensemble AI to decrease false positives. Our data support the sparse annotation technique as an efficient technique for reducing the time needed to establish the ground truth.

A multi-module structure labelled molecular network orients the chemical profiles of traditional Chinese medicine prescriptions: Xiaoyao San, as an example

Ultra-high-performance liquid chromatography-mass spectrometry (UHPLC-MS) technology has emerged as a crucial tool for identifying components in traditional Chinese medicine (TCM). However, the characterization of the chemical profiles of TCM prescriptions (TCMPs) which often consist of multiple herbal medicines and contain diverse structural types, presents several challenges, such as component overlapping and time-consuming. In this study, a novel strategy known as the multi-module structure labelled molecular network (MSLMN), which integrates molecular networking, database annotation, and cluster analysis techniques, has been successfully proposed, which facilitates the identification of chemical constituents by leveraging a high-structural similarity ion list derived from the MSLMN. It has been effectively applied to analyze the chemical profile of Xiaoyao San (XYS), a classical TCMP. Through the MSLMN method, a total of 302 chemical constituents were identified, covering nine structural types in XYS. Furthermore, a validated and quantitative analytical method using UHPLC-QqQ-MS/MS technology was developed for 31 identified chemicals, encompassing all eight herbal medicines present in XYS, and the developed analytical approach was applied to investigate the content distribution across 40 different batches of commercially available XYS. In total, the proposed strategy has practical significance for improving the insight into the chemical profile of XYS and serves as a valuable approach for handling complex system data based on UHPLC-MS, particularly for TCMPs.

Enhancing health-care data integration via automated semantic mapping

PurposeThis paper aims to delve into the complexities of terminology mapping and annotation, particularly within the context of the COVID-19 pandemic. It underscores the criticality of harmonizing clinical knowledge organization systems (KOS) through a cohesive clinical knowledge representation approach. Central to the study is the pursuit of a novel method for integrating emerging COVID-19-specific vocabularies with existing systems, focusing on simplicity, adaptability and minimal human intervention.Design/methodology/approachA design science research (DSR) methodology is used to guide the development of a terminology mapping and annotation workflow. The KNIME data analytics platform is used to implement and test the mapping and annotation techniques, leveraging its powerful data processing and analytics capabilities. The study incorporates specific ontologies relevant to COVID-19, evaluates mapping accuracy and tests performance against a gold standard.FindingsThe study demonstrates the potential of the developed solution to map and annotate specific KOS efficiently. This method effectively addresses the limitations of previous approaches by providing a user-friendly interface and streamlined process that minimizes the need for human intervention. Additionally, the paper proposes a reusable workflow tool that can streamline the mapping process. It offers insights into semantic interoperability issues in health care as well as recommendations for work in this space.Originality/valueThe originality of this study lies in its use of the KNIME data analytics platform to address the unique challenges posed by the COVID-19 pandemic in terminology mapping and annotation. The novel workflow developed in this study addresses known challenges by combining mapping and annotation processes specifically for COVID-19-related vocabularies. The use of DSR methodology and relevant ontologies with the KNIME tool further contribute to the study’s originality, setting it apart from previous research in the terminology mapping and annotation field.

A large-scale phylogeny-guided analysis of pseudogenes in Pseudomonas aeruginosa bacterium.

Pseudogenes, once considered "junk DNA" based on the incorrect assumption that the absence of full coding potential means a complete lack of functionality, have recently become a subject of significant interest in the scientific community. Concurrently, it is widely assumed that bacterial genomes are compact and have a high density of coding genes with little room for non-coding genes, including pseudogenes. A key aspect of genome annotation is the correct identification of genes and the distinction between coding genes and pseudogenes, as it directly impacts functional and comparative genomics studies. In this study, we analyzed the genomic data of 4,699 strains of the bacterium Pseudomonas aeruginosa (P. aeruginosa) as they exhibit high variability in the number of annotated pseudogenes. In particular, we looked for correlations between the number of pseudogenes and other genomic and meta-features of the strains. We identified clusters of orthologous genes and pseudogenes and compared cluster size distributions and length homogeneity within clusters. We then mapped and examined orthology relationships between genes and pseudogenes. Additionally, we generated a phylogenetic tree of the strains and found that phylogenetically related strains are more homogeneous in the number of pseudogenes and share a significant amount of pseudogenes. Finally, we delved into clusters of orthologous genes and pseudogenes and quantified their phylogenetic neighborhood, classifying pseudogenes into evolutionary preserved pseudogenes, mis-annotated pseudogenes, or pseudogenes formed by failed horizontal transfer events. This in-depth study provides important insights that can be incorporated into pseudogene annotation pipelines in the future. IMPORTANCE Accurate annotation of genes and pseudogenes is vital for comparative genomics analysis. Recent studies have shown that bacterial pseudogenes have an important role in regulatory processes and can provide insight into the evolutionary history of homologous genes or the genome as a whole. Due to pseudogenes' nature as non-functional genes, there is no commonly accepted definition of a pseudogene, which poses difficulties in verifying the annotation through experimental methods and resolving discrepancies among different annotation techniques. Our study introduces an in-depth analysis of annotated genes and pseudogenes and insights that can be incorporated into improved pseudogene annotation pipelines in the future.

Methods for cell-type annotation on scRNA-seq data: A recent overview.

The evolution of single-cell technology is ongoing, continually generating massive amounts of data that reveal many mysteries surrounding intricate diseases. However, their drawbacks continue to constrain us. Among these, annotating cell types in single-cell gene expressions pose a substantial challenge, despite the myriad of tools at our disposal. The rapid growth in data, resources, and tools has consequently brought about significant alterations in this area over the years. In our study, we spotlight all note-worthy cell type annotation techniques developed over the past four years. We provide an overview of the latest trends in this field, showcasing the most advanced methods in taxonomy. Our research underscores the demand for additional tools that incorporate a biological context and also predicts that the rising trend of graph neural network approaches will likely lead this research field in the coming years.

Development and validation of a differentiation-related signature based on single-cell RNA sequencing data of immune cells in spinal cord injury

BackgroundAfter spinal cord injury (SCI), the native immune surveillance function of the central nervous system is activated, resulting in a substantial infiltration of immune cells into the affected tissue. While numerous studies have explored the transcriptome data following SCI and revealed certain diagnostic biomarkers, there remains a paucity of research pertaining the identification of immune subtypes and molecular markers related to the immune system post-spinal cord injury using single-cell sequencing data of immune cells. MethodsThe researchers conducted an analysis of spinal cord samples obtained at three time points (3,10, and 21 days) following SCI using the GSE159638 dataset. The SCI subsets were delineated through pseudo-time analysis, and differentiation related genes were identified after principal component analysis (PCA), cell clustering, and annotation techniques. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were employed to assess the differentiation-related genes (DRGs) across different subsets. The molecular subtypes of SCI were determined using consensus clustering analysis. To further explore and validate the correlation between the molecular subtypes and the immune microenvironment, the CIBERSORT algorithm was employed. High-value diagnostic gene markers were identified using LASSO regression, and their diagnostic sensitivity was assessed using receiver operating characteristic curves (ROC) and quantitative real-time polymerase chain reaction (qRT-PCR). ResultsThree SCI subsets were obtained, and differentiation-related genes were characterized. Within these subsets, two distinct molecular subtypes, namely C1 and C2, were identified. These subtypes demonstrated significant variations in terms of immune cell infiltration levels and the expression of immune checkpoint genes. Through further analysis, three candidate biomarkers (C1qa, Lgals3 and Cd63) were identified and subsequently validated. ConclusionsOur study revealed a diverse immune microenvironment in SCI samples, highlighting the potential significance of C1qa, Lgals3 and Cd63 as immune biomarkers for diagnosing SCI. Moreover, the identification of immune checkpoints corresponding to the two molecular subtypes suggests their potential as targets for immunotherapy to enhance SCI repair in future interventions.

Sweetgum Leaf Spot Image Segmentation and Grading Detection Based on an Improved DeeplabV3+ Network

Leaf spot disease and brown spot disease are common diseases affecting maple leaves. Accurate and efficient detection of these diseases is crucial for maintaining the photosynthetic efficiency and growth quality of maple leaves. However, existing segmentation methods for plant diseases often fail to accurately and rapidly detect disease areas on plant leaves. This paper presents a novel solution to accurately and efficiently detect common diseases in maple leaves. We propose a deep learning approach based on an enhanced version of DeepLabV3+ specifically designed for detecting common diseases in maple leaves. To construct the maple leaf spot dataset, we employed image annotation and data enhancement techniques. Our method incorporates the CBAM-FF module to fuse gradual features and deep features, enhancing the detection performance. Furthermore, we leverage the SANet attention mechanism to improve the feature extraction capabilities of the MobileNetV2 backbone network for spot features. The utilization of the focal loss function further enhances the detection accuracy of the affected areas. Experimental results demonstrate the effectiveness of our improved algorithm, achieving a mean intersection over union (MIoU) of 90.23% and a mean pixel accuracy (MPA) of 94.75%. Notably, our method outperforms traditional semantic segmentation methods commonly used for plant diseases, such as DeeplabV3+, Unet, Segnet, and others. The proposed approach significantly enhances the segmentation performance for detecting diseased spots on Liquidambar formosana leaves. Additionally, based on pixel statistics, the segmented lesion image is graded for accurate detection.

In-the-Wild Affect Analysis of Children with ASD Using Heart Rate.

Recognizing the affective state of children with autism spectrum disorder (ASD) in real-world settings poses challenges due to the varying head poses, illumination levels, occlusion and a lack of datasets annotated with emotions in in-the-wild scenarios. Understanding the emotional state of children with ASD is crucial for providing personalized interventions and support. Existing methods often rely on controlled lab environments, limiting their applicability to real-world scenarios. Hence, a framework that enables the recognition of affective states in children with ASD in uncontrolled settings is needed. This paper presents a framework for recognizing the affective state of children with ASD in an in-the-wild setting using heart rate (HR) information. More specifically, an algorithm is developed that can classify a participant's emotion as positive, negative, or neutral by analyzing the heart rate signal acquired from a smartwatch. The heart rate data are obtained in real time using a smartwatch application while the child learns to code a robot and interacts with an avatar. The avatar assists the child in developing communication skills and programming the robot. In this paper, we also present a semi-automated annotation technique based on facial expression recognition for the heart rate data. The HR signal is analyzed to extract features that capture the emotional state of the child. Additionally, in this paper, the performance of a raw HR-signal-based emotion classification algorithm is compared with a classification approach based on features extracted from HR signals using discrete wavelet transform (DWT). The experimental results demonstrate that the proposed method achieves comparable performance to state-of-the-art HR-based emotion recognition techniques, despite being conducted in an uncontrolled setting rather than a controlled lab environment. The framework presented in this paper contributes to the real-world affect analysis of children with ASD using HR information. By enabling emotion recognition in uncontrolled settings, this approach has the potential to improve the monitoring and understanding of the emotional well-being of children with ASD in their daily lives.

Method and Subject

Research activities in the social sciences and humanities have traditionally conceived digital outputs in terms of databases and websites. Practically, most of these are still implemented using SQL, table-based, data structures and synchronic web technologies. However, the constant evolution of software means that browser functionality and security arrangements are constantly changing. As a result, few digital research outputs, except published literature, remain accessible for more than a few years. Recent progress with research data infrastructures is described, which has the potential to improve the sustainability of research investments. This article presents new standards-based annotation techniques, developed in the biodiversity community, which have been applied to global history research questions. Open repository software platforms supporting this ‘scientific treatment’ approach can now generate technology-independent data resources – supporting long-term reuse by the global community. Promoting institutional change to adopt these developments is discussed, so that costs of data stewardship can be made forecast-able.

Deep learning based active learning technique for data annotation and improve the overall performance of classification models

Deep learning (DL) models achieved a lot of success due to the availability of labeled training data. In contrast, labeling a huge amount of data by a human is a time-consuming and expensive solution. Active Learning (AL) efficiently addresses the issue of labeled data collection at a low cost by picking the most useful samples from a large number of unlabeled datasets. However, current AL techniques largely depend on regular human involvement to annotate the most uncertain/informative samples in the collection. Therefore, we introduce a Deep Learning Based Active Learning (DLBAL) method that may incrementally learn from a small number of annotated training samples to build an effective classifier with optimum feature representation. The proposed DLBAL method improves the contemporary AL approaches in two steps. First, EfficientNet-B0 networks are integrated into AL. The incrementally annotated useful samples may be used to update both the representation of the feature and the classification algorithm concurrently. Second, we provide an effective technique for sample selection that improves the performance of the classifier with fewer manual annotations at a lower cost. In contrast to previous approaches that emphasize only uncertain instances with low prediction confidence, we obtain a significant number of instances with high-confidence from the unannotated set for pattern learning. Particularly, these instances are randomly picked and assigned labels to them continuously. Detailed experiments are conducted to show that the proposed DLBAL system obtained reasonable results on two difficult image classification datasets, namely Cross-Age Celebrity face recognition Dataset (CACD) and Caltech-256.

Cross-platform dataset of multiplex fluorescent cellular object image annotations

Defining cellular and subcellular structures in images, referred to as cell segmentation, is an outstanding obstacle to scalable single-cell analysis of multiplex imaging data. While advances in machine learning-based segmentation have led to potentially robust solutions, such algorithms typically rely on large amounts of example annotations, known as training data. Datasets consisting of annotations which are thoroughly assessed for quality are rarely released to the public. As a result, there is a lack of widely available, annotated data suitable for benchmarking and algorithm development. To address this unmet need, we release 105,774 primarily oncological cellular annotations concentrating on tumor and immune cells using over 40 antibody markers spanning three fluorescent imaging platforms, over a dozen tissue types and across various cellular morphologies. We use readily available annotation techniques to provide a modifiable community data set with the goal of advancing cellular segmentation for the greater imaging community.

İstanbul Street Rhythms: A Field Guide to Short Expressive Ensembles

This article presents the field guide as an alternate format to the anthology in documenting short literary expressions. Anthologies often isolate, collect, and remove literary forms from their habitats by representing them in a book, altering the context by which short literary expressions can be interpreted. After the Gezi Park protests in İstanbul in 2013, several print and electronic anthologies documented public expressions, yet these books offer no or only vague mention of location. Through Henri Lefebvre’s Rhythmanalysis: Space Time and Everyday Life, I emphasize the rhythms of the street that are interrupted by public expressions. After a review of Lefebvre’s work on rhythms, I offer digital annotation techniques to map the material and medium of public inscriptions of one neighborhood of Kadıköy, İstanbul called Rasimpaşa, a district historically called “Yeldeğirmeni” or “The Windmill.” The focus on public graffiti, street art, posters, signs, stickers, or other verbal traces in public space allows for an expansion of the concept of “short literature” to include their lived habitats as rhythmic ensembles.

Automatic Detection, Classification, and Grading of Lumbar Intervertebral Disc Degeneration Using an Artificial Neural Network Model.

Intervertebral disc degeneration (IDD) is a common cause of symptomatic axial low back pain. Magnetic resonance imaging (MRI) is currently the standard for the investigation and diagnosis of IDD. Deep learning artificial intelligence models represent a potential tool for rapidly and automatically detecting and visualizing IDD. This study investigated the use of deep convolutional neural networks (CNNs) for the detection, classification, and grading of IDD. Sagittal images of 1000 IDD T2-weighted MRI images from 515 adult patients with symptomatic low back pain were separated into 800 MRI images using annotation techniques to create a training dataset (80%) and 200 MRI images to create a test dataset (20%). The training dataset was cleaned, labeled, and annotated by a radiologist. All lumbar discs were classified for disc degeneration based on the Pfirrmann grading system. The deep learning CNN model was used for training in detecting and grading IDD. The results of the training with the CNN model were verified by testing the grading of the dataset using an automatic model. The training dataset of the sagittal intervertebral disc lumbar MRI images found 220 IDDs of grade I, 530 of grade II, 170 of grade III, 160 of grade IV, and 20 of grade V. The deep CNN model was able to detect and classify lumbar IDD with an accuracy of more than 95%. The deep CNN model can reliably automatically grade routine T2-weighted MRIs using the Pfirrmann grading system, providing a quick and efficient method for lumbar IDD classification.

Enhancing Image Annotation Technique of Fruit Classification Using a Deep Learning Approach

An accurate image retrieval technique is required due to the rapidly increasing number of images. It is important to implement image annotation techniques that are fast, simple, and, most importantly, automatically annotate. Image annotation has recently received much attention due to the massive rise in image data volume. Focusing on the agriculture field, this study implements automatic image annotation, namely, a repetitive annotation task technique, to classify the ripeness of oil palm fruit and recognize a variety of fruits. This approach assists farmers to enhance the classification of fruit methods and increase their production. This study proposes simple and effective models using a deep learning approach with You Only Look Once (YOLO) versions. The models were developed through transfer learning where the dataset was trained with 100 images of oil fruit palm and 400 images of a variety of fruit in RGB images. Model performance and accuracy of automatically annotating the images with 3500 fruits were examined. The results show that the annotation technique successfully annotated a large number of images accurately. The mAP result achieved for oil palm fruit was 98.7% and the variety of fruit was 99.5%.

Integrating structure annotation and machine learning approaches to develop graphene toxicity models

Modern nanotechnology provides efficient and cost-effective nanomaterials (NMs). The increasing usage of NMs brings great concerns regarding nanotoxicity in humans. Traditional animal testing of nanotoxicity is expensive and time-consuming. Modeling studies using machine learning (ML) approaches are promising alternatives to direct evaluation of nanotoxicity based on nanostructure features. However, NMs, including two-dimensional nanomaterials (2DNMs) such as graphenes, have complex structures making them difficult to annotate and quantify the nanostructures for modeling purposes. To address this issue, we constructed a virtual graphenes library using nanostructure annotation techniques. The irregular graphene structures were generated by modifying virtual nanosheets. The nanostructures were digitalized from the annotated graphenes. Based on the annotated nanostructures, geometrical nanodescriptors were computed using Delaunay tessellation and used for ML modeling. Partial least squares regression (PLSR) models for the graphenes were built and validated using a leave-one-out cross-validation (LOOCV) procedure. The resulted models showed good predictivity in four toxicity-related endpoints with the coefficient of determination (R2) ranging from 0.558 to 0.822. This study provides a novel nanostructure annotation strategy that can be applied to generate high-quality nanodescriptors for ML model developments, which can be widely applied to nanoinformatics studies of graphenes and other NMs.

ETCNN: Extra Tree and Convolutional Neural Network-based Ensemble Model for COVID-19 Tweets Sentiment Classification.

Pandemics influence people negatively and people experience fear and disappointment. With the global outspread of COVID-19, the sentiments of the general public are substantially influenced, and analyzing their sentiments could help to devise corresponding policies to alleviate negative sentiments. Often the data collected from social media platforms is unstructured leading to low classification accuracy. This study brings forward an ensemble model where the benefits of handcrafted features and automatic feature extraction are combined by machine learning and deep learning models. Unstructured data is obtained, preprocessed, and annotated using TextBlob and VADER before training machine learning models. Similarly, the efficiency of Word2Vec, TF, and TF-IDF features is also analyzed. Results reveal the better performance of the extra tree classifier when trained with TF-IDF features from TextBlob annotated data. Overall, machine learning models perform better with TF-IDF and TextBlob. The proposed model obtains superior performance using both annotation techniques with 0.97 and 0.95 scores of accuracy using TextBlob and VADER respectively with Word2Vec features. Results reveal that use of machine learning and deep learning models together with a voting criterion tends to yield better results than other machine learning models. Analysis of sentiments indicates that predominantly people possess negative sentiments regarding COVID-19.

A Survey on Physiological Signal-Based Emotion Recognition.

Physiological signals are the most reliable form of signals for emotion recognition, as they cannot be controlled deliberately by the subject. Existing review papers on emotion recognition based on physiological signals surveyed only the regular steps involved in the workflow of emotion recognition such as pre-processing, feature extraction, and classification. While these are important steps, such steps are required for any signal processing application. Emotion recognition poses its own set of challenges that are very important to address for a robust system. Thus, to bridge the gap in the existing literature, in this paper, we review the effect of inter-subject data variance on emotion recognition, important data annotation techniques for emotion recognition and their comparison, data pre-processing techniques for each physiological signal, data splitting techniques for improving the generalization of emotion recognition models and different multimodal fusion techniques and their comparison. Finally, we discuss key challenges and future directions in this field.

MEGACOSM – DETECTION AND CLASSIFICATION OF ASTRONOMICAL OBJECTS

The concept of existence started with the bigbang theory, which was a phenomenon where multiple objects collided to create multiple other things. This creation is said to be increasing at the rate of the cosmic acceleration due to which multiple new objects came into existence called as the astronomical objects. Space object (SO) detection, classification, and characterization are significant challenges in many research fields. In recent years, deep learning and other forms of artificial intelligence (AI) have drawn the attention of many astronomers and academics. Megacosm is project that is used for the classification and the identification of those newly created celestial objects. It works on the process of manually training the model with the data annotation techniques and the dataset is enhanced using the data augmentation technique. It uses YOLO as the core algorithm and also deep learning concepts like CNN (Convolution Neural Network) to predict results. It gives the output as a bounding box around the detected object along with the accuracy of that prediction.

Advanced Technology in Agriculture Industry by Implementing Image Annotation Technique and Deep Learning Approach: A Review

The implementation of intelligent technology in agriculture is seriously investigated as a way to increase agriculture production while reducing the amount of human labor. In agriculture, recent technology has seen image annotation utilizing deep learning techniques. Due to the rapid development of image data, image annotation has gained a lot of attention. The use of deep learning in image annotation can extract features from images and has been shown to analyze enormous amounts of data successfully. Deep learning is a type of machine learning method inspired by the structure of the human brain and based on artificial neural network concepts. Through training phases that can label a massive amount of data and connect them up with their corresponding characteristics, deep learning can conclude unlabeled data in image processing. For complicated and ambiguous situations, deep learning technology provides accurate predictions. This technology strives to improve productivity, quality and economy and minimize deficiency rates in the agriculture industry. As a result, this article discusses the application of image annotation in the agriculture industry utilizing several deep learning approaches. Various types of annotations that were used to train the images are presented. Recent publications have been reviewed on the basis of their application of deep learning with current advancement technology. Plant recognition, disease detection, counting, classification and yield estimation are among the many advancements of deep learning architecture employed in many applications in agriculture that are thoroughly investigated. Furthermore, this review helps to assist researchers to gain a deeper understanding and future application of deep learning in agriculture. According to all of the articles, the deep learning technique has successfully created significant accuracy and prediction in the model utilized. Finally, the existing challenges and future promises of deep learning in agriculture are discussed.