Image Classification Techniques Research Articles

Hash-Based Deep Learning Approach for Remote Sensing Satellite Imagery Detection

Ship detection plays a crucial role in marine security in remote sensing imagery. This paper discusses about a deep learning approach to detect the ships from satellite imagery. The model developed in this work achieves integrity by the inclusion of hashing. This model employs a supervised image classification technique to classify images, followed by object detection using You Only Look Once version 3 (YOLOv3) to extract features from deep CNN. Semantic segmentation and image segmentation is done to identify object category of each pixel using class labels. Then, the concept of hashing using SHA-256 is applied in conjunction with the ship count and location of bounding box in satellite image. The proposed model is tested on a Kaggle Ships dataset, which consists of 231,722 images. A total of 70% of this data is used for training, and the 30% is used for testing. To add security to images with detected ships, the model is enhanced by hashing using SHA-256 algorithm. Using SHA-256, which is a one-way hash, the data are split up into blocks of 64 bytes. The input data to the hash function are both the ship count and bounding box location. The proposed model achieves integrity by using SHA-256. This model allows secure transmission of highly confidential images that are tamper-proof.

Early Detection of Medical Image Analysis by Using Machine Learning Method.

We develop effective medical image classification techniques, with an emphasis on histopathology and magnetic resonance imaging (MRI). The trainer utilized the curriculum as a starting point for a set of data and a restricted number of samples, and we used it as a starting point for a set of data. As calibrating a machine learning model is difficult, we used alternative methods as unsupervised feature extracts or weight-conditioning factors for identifying pathological histology pictures. As a result, the pretrained models will be trained on 3-channel RGB pictures, while the MRI sample has more slices. To alter the working model using the MRI data, the convolutional neural network (CNN) must be fine-tuned. Pretrained models are placed and then used as feature snippets. However, there is a scarcity of well-done medical photos, making training machine learning models a difficult endeavor to begin with. In any case, data augmentation aids in the generation of sufficient training samples; however, it is unclear if data augmentation aids in the prediction of unknown data samples. As a result, we fine-tuned machine learning models without using any additional data. Furthermore, rather than utilizing a standard machine learning classifier for the MRI data, we created a unique CNN that uses both 3D shear descriptors and deep features as input. This custom network identifies the MRI sample after processing our representation of the characteristics from beginning to end. On the hidden MRI dataset, our bespoke CNN outperforms traditional machine learning. Our CNN model is less prone to overfitting as a result of this. Furthermore, we have given cutting-edge outcomes employing machine learning.

A Simple and Efficient Deep Learning-Based Framework for Automatic Fruit Recognition.

Accurate detection and recognition of various kinds of fruits and vegetables by using the artificial intelligence (AI) approach always remain a challenging task due to similarity between various types of fruits and challenging environments such as lighting and background variations. Therefore, developing and exploring an expert system for automatic fruits' recognition is getting more and more important after many successful approaches; however, this technology is still far from being mature. The deep learning-based models have emerged as state-of-the-art techniques for image segmentation and classification and have a lot of promise in challenging domains such as agriculture, where they can deal with the large variability in data better than classical computer vision methods. In this study, we proposed a deep learning-based framework to detect and recognize fruits and vegetables automatically with difficult real-world scenarios. The proposed method might be helpful for the fruit sellers to identify and differentiate various kinds of fruits and vegetables that have similarities. The proposed method has applied deep convolutional neural network (DCNN) to the undertakings of distinguishing natural fruit images of the Gilgit-Baltistan (GB) region as this area is famous for fruits' production in Pakistan as well as in the world. The experimental outcomes demonstrate that the suggested deep learning algorithm has the effective capability of automatically recognizing the fruit with high accuracy of 96%. This high accuracy exhibits that the proposed approach can meet world application requirements.

Detecting breast cancer using novel mask R‐CNN techniques

AbstractBreast cancer is more common in women and the mortality rate also increases in recent days. Early detection of breast cancer is reduces the severity of the disease to some extent. Various image processing and classification techniques imposed on a particular image to detect and diagnose the breast cancer clearly. For early detection, an efficient methodology is needed. To reduce death rate, accurate discovery of the disease should be efficiently implemented. The mask R‐CNN is used for segmentation to identify abnormalities and ensemble CNN is used to classify the benign and malignant tumour from the given mammographic breast cancer image. The input is fed into the enhanced fuzzy based median filter which is used to remove speckle noises which in turn increases the clarity of images. The noise‐free images are fed into the proposed mask R‐CNN for segmentation. It is an effective framework for segmentation of medical images. The classification process is then given into ensemble based CNN classifier for prediction of mammogram images. The ensemble based CNN is proposed for the classification of mammograms as benign and malignant. The proposed method focus on the optimization which is performed to minimize the memory requirements and running time while maintaining the high performance of the classifier. The performance metrics such as accuracy, precision, f‐measure, and recall is evaluated to measure the efficacy of the proposed design. Using MIAS and DDSM dataset the performance metrics has been evaluated and compared with existing approaches. The results shows that the performance of the proposed method shows higher efficiency.

Explainable image classification with evidence counterfactual

The complexity of state-of-the-art modeling techniques for image classification impedes the ability to explain model predictions in an interpretable way. A counterfactual explanation highlights the parts of an image which, when removed, would change the predicted class. Both legal scholars and data scientists are increasingly turning to counterfactual explanations as these provide a high degree of human interpretability, reveal what minimal information needs to be changed in order to come to a different prediction and do not require the prediction model to be disclosed. Our literature review shows that existing counterfactual methods for image classification have strong requirements regarding access to the training data and the model internals, which often are unrealistic. Therefore, SEDC is introduced as a model-agnostic instance-level explanation method for image classification that does not need access to the training data. As image classification tasks are typically multiclass problems, an additional contribution is the introduction of the SEDC-T method that allows specifying a target counterfactual class. These methods are experimentally tested on ImageNet data, and with concrete examples, we illustrate how the resulting explanations can give insights in model decisions. Moreover, SEDC is benchmarked against existing model-agnostic explanation methods, demonstrating stability of results, computational efficiency and the counterfactual nature of the explanations.

An outlook: machine learning in hyperspectral image classification and dimensionality reduction techniques

Hyperspectral imaging is used in a wide range of applications. When used in remote sensing, satellites and aircraft are employed to collect the images, which are used in agriculture, environmental monitoring, urban planning and defence. The exact classification of ground features in the images is a significant research issue and is currently receiving greater attention. Moreover, these images have a large spectral dimensionality, which adds computational complexity and affects classification precision. To handle these issues, dimensionality reduction is an essential step that improves the performance of classifiers. In the classification process, several strategies have produced good classification results. Of these, machine learning techniques are the most powerful approaches. As a result, this paper reviews three different types of hyperspectral image machine learning classification methods: cluster analysis, supervised and semi-supervised classification. Moreover, this paper shows the effectiveness of all these techniques for hyperspectral image classification and dimensionality reduction. Furthermore, this review will assist as a reference for future research to improve the classification and dimensionality reduction approaches.

A semi‐automatic workflow for plot boundary extraction of irregularly sized and spaced field plots from UAV imagery

AbstractAdvances in high‐throughput platforms such as UAVs (unoccupied aerial vehicles) facilitate rapid image‐based phenotypic data acquisition. However, existing plot‐level data extraction methods are unreliable if field plots differ in size and spacing, as often occurs in early‐generation plant breeding trials. To overcome the limitations of conventional plot extraction techniques, a combinational approach with both field‐map information and image classification techniques can be used to optimize plot extraction. The objective of this study was to develop a plot boundary extraction workflow for irregularly sized and spaced field plots from UAV imagery using plot spacing data and vegetation index‐based classifiers. An herbicide screening experiment consisting of three replications of 780 lentil (Lens culinaris Medik.) populations was foliar sprayed with saflufenacil. Aerial image acquisition was conducted during the peak vegetation stage using a RedEdge multispectral camera. A semi‐automatic workflow was compiled in eCognition software to extract lentil plot boundaries. Normalized difference vegetation index (NDVI) was calculated to locate the plots with vegetation and those with low NDVI or no vegetation, and pixel resizing based on plot size and orientation was used to draw the plot boundary. The extraction results showed a precise estimation of plot boundary for all the plots with a wide range of herbicide damage, including the plots with complete loss of vegetation. By using a simple convolutional filter (line filter), image thresholding, and pixel resizing, this approach avoided the use of complex algorithm‐based methodologies. Results suggest that this workflow can be extended to a wide range of phenotyping studies.

Identification of Foliar Disease Regions on Corn Leaves Using SLIC Segmentation and Deep Learning Under Uniform Background and Field Conditions

Plant diseases lead to severe losses in crop yield worldwide. The conventional approach for diagnosing diseases relies on manual scouting. In recent years, advances in convolutional neural networks have motivated the use of deep learning-based computer vision for automatically identifying plant diseases. Although image classification techniques are commonly used for analyzing agricultural data, their use for accurately identifying diseased regions corresponding to different disease types on individual plant leaves is limited. In this study, Simple Linear Iterative Clustering (SLIC) segmentation was used on corn leaf images from the PlantVillage and CD&S datasets to create super-pixels, a cluster of pixels representing a region of interest on a corn leaf. The VGG16, ResNet50, DenseNet121, Xception, and InceptionV3, pre-trained deep learning models were utilized to identify diseased regions corresponding to five super-pixel classes (healthy, northern leaf blight (NLB), gray leaf spot (GLS), common rust, and background) for the PlantVillage dataset and four super-pixel classes (NLB, GLS, northern leaf spot, and background) for the CD&S dataset, on corn leaves. After setting the spatial proximity value (sigma) for SLIC segmentation to five, a total of 100 models were trained by using different numbers of segments per image (five and fifteen) in the SLIC algorithm for both datasets and choosing training: testing split ratios of 90:10, 80:20, 70:30, 60:40, and 50:50 for each of the five models. The highest overall testing accuracy of 97.77% was observed after training the DenseNet121 model to identify super-pixels created from the CD&S dataset, belonging to the four classes created using a sigma value of five, five segments per image, and an 80:20 training: testing split ratio. Web and mobile applications were developed to identify diseased regions on corn leaves using the best deep learning model as the classifier. The results suggest that SLIC segmentation on corn leaf images helps accurately identify diseased regions. This research demonstrates the potential of image-based scouting as an efficient alternative to manual scouting for disease monitoring.

A Deep Learning Approach to Manage and Reduce Plastic Waste in the Oceans

The accumulation of plastic objects in the Earth’s environment will adversely affect wildlife, wildlife habitat, and humans. The huge amount of unrecycled plastic ends up in landfill and thrown into unregulated dump sites. In many cases, specifically in the developing countries, plastic waste is thrown into rivers, streams and oceans. In this work, we employed the power of deep learning techniques in image processing and classification to recognize plastic waste. Our work aims to identify plastic texture and plastic objects in images in order to reduce plastic waste in the oceans, and facilitate waste management. For this, we use transfer learning in two ways: in the first one, a pre-trained CNN model on ImageNet is used as a feature extractor, then an SVM classifier for classification, the second strategy is based on fine tuning the pre-trained CNN model. Our approach was trained and tested using two (02) challenging datasets one is a texture recognition dataset and the other is for object detection, and achieves very satisfactory results using two (02) deep learning strategies.

ResNet50-Based Effective Model for Breast Cancer Classification Using Histopathology Images

Breast cancer is considered an immense threat and one of the leading causes of mortality in females. It is curable only when detected at an early stage. A standard cancer diagnosis approach involves detection of cancer-related anomalies in tumour histopathology images. Detection depends on the accurate identification of the landmarks in the visual artefacts present in the slide images. Researchers are continuously striving to develop automatic machine-learning algorithms for processing medical images to assist in tumour detection. Nowadays, computer-based automated systems play an important role in cancer image analysis and help healthcare experts make rapid and correct inferences about the type of cancer. This study proposes an effective convolutional neural network-based (CNN-based) model that exploits the transfer-learning technique for automatic image classification between malignant and benign tumour, using histopathology images. Resnet50 architecture has been trained on new dataset for feature extraction, and fully connected layers have been fine-tuned for achieving highest training, validation and test accuracies. The result illustrated state-of-the-art performance of the proposed model with highest training, validation and test accuracies as 99.70%, 99.24% and 99.24%, respectively. Classification accuracy is increased by 0.66% and 0.2% when compared with similar recent studies on training and test data results. Average precision and F1 score have also improved, and receiver operating characteristic (RoC) area has been achieved to 99.1%. Thus, a reliable, accurate and consistent CNN model based on pre-built Resnet50 architecture has been developed.

Efficient Data Augmentation Techniques for Improved Classification in Limited Data Set of Oral Squamous Cell Carcinoma

Deep Learning (DL) techniques as a subfield of data science are getting overwhelming attention mainly because of their ability to understand the underlying pattern of data in making classifications. These techniques require a considerable amount of data to efficiently train the DL models. Generally, when the data size is larger, the DL models perform better. However, it is not possible to have a considerable amount of data in different domains such as healthcare. In healthcare, it is impossible to have a substantial amount of data to solve medical problems using Artificial Intelligence, mainly due to ethical issues and the privacy of patients. To solve this problem of small dataset, different techniques of data augmentation are used that can increase the size of the training set. However, these techniques only change the shape of the image and hence the classification model does not increase accuracy. Generative Adversarial Networks (GANs) are very powerful techniques to augment training data as new samples are created. This technique helps the classification models to increase their accuracy. In this paper, we have investigated augmentation techniques in healthcare image classification. The objective of this research paper is to develop a novel augmentation technique that can increase the size of the training set, to enable deep learning techniques to achieve higher accuracy. We have compared the performance of the image classifiers using the standard augmentation technique and GANs. Our results demonstrate that GANs increase the training data, and eventually, the classifier achieves an accuracy of 90% compared to standard data augmentation techniques, which achieve an accuracy of up to 70%. Other advanced CNN models are also tested and have demonstrated that more deep architectures can achieve more than 98% accuracy for making classification on Oral Squamous Cell Carcinoma.

Optimal IoT Based Improved Deep Learning Model for Medical Image Classification

Recently medical image classification plays a vital role in medical image retrieval and computer-aided diagnosis system. Despite deep learning has proved to be superior to previous approaches that depend on handcrafted features; it remains difficult to implement because of the high intra-class variance and inter-class similarity generated by the wide range of imaging modalities and clinical diseases. The Internet of Things (IoT) in healthcare systems is quickly becoming a viable alternative for delivering high-quality medical treatment in today’s e-healthcare systems. In recent years, the Internet of Things (IoT) has been identified as one of the most interesting research subjects in the field of health care, notably in the field of medical image processing. For medical picture analysis, researchers used a combination of machine and deep learning techniques as well as artificial intelligence. These newly discovered approaches are employed to determine diseases, which may aid medical specialists in disease diagnosis at an earlier stage, giving precise, reliable, efficient, and timely results, and lowering death rates. Based on this insight, a novel optimal IoT-based improved deep learning model named optimization-driven deep belief neural network (ODBNN) is proposed in this article. In context, primarily image quality enhancement procedures like noise removal and contrast normalization are employed. Then the pre-processed image is subjected to feature extraction techniques in which intensity histogram, an average pixel of RGB channels, first-order statistics, Grey Level Co-Occurrence Matrix, Discrete Wavelet Transform, and Local Binary Pattern measures are extracted. After extracting these sets of features, the May Fly optimization technique is adopted to select the most relevant features. The selected features are fed into the proposed classification algorithm in terms of classifying similar input images into similar classes. The proposed model is evaluated in terms of accuracy, precision, recall, and f-measure. The investigation evident the performance of incorporating optimization techniques for medical image classification is better than conventional techniques.

Deep Encoder With Kernel-Wise Taylor Series for Hyperspectral Image Classification

Deep learning is a popular and effective technique for the hyperspectral image (HSI) classification. Current deep learning-based methods have numerous free parameters to be trained. They may be unavailable once lacking training samples. In addition, these approaches only use the features from the deepest layer and exclude shallow features, which is a kind of information loss. To remedy such deficiencies, in this work, we construct a novel deep encoder with kernel-wise Taylor series (EKTS) for the HSI classification. More specifically, we introduce the Taylor series to approximate the role of deep networks for feature extraction. Because the original Taylor series is linear, the kernel theory is used to build the kernel-wise Taylor series to encode the HSI data and extract deep nonlinear features. Furthermore, an alternating iterative optimization strategy is developed to obtain the outputs of all layers of the built deep encoder. Subsequently, we stack the outputs of all layers to obtain the final features that integrate both shallow and deep features. At last, the support vector machine (SVM) is adopted to deal with the obtained final features so that label results can be predicted. Unlike current deep learning methods, our EKTS has no free parameters to be trained and combines the advantages of both shallow and deep features to predict labels. Sufficient experimental analysis has been performed to verify the greater classification performance of our EKTS method compared with many state-of-the-art approaches.

Low Complexity Encoder with Multilabel Classification and Image Captioning燤odel

Due to the advanced development in the multimedia-on-demand traffic in different forms of audio, video, and images, has extremely moved on the vision of the Internet of Things (IoT) from scalar to Internet of Multimedia Things (IoMT). Since Unmanned Aerial Vehicles (UAVs) generates a massive quantity of the multimedia data, it becomes a part of IoMT, which are commonly employed in diverse application areas, especially for capturing remote sensing (RS) images. At the same time, the interpretation of the captured RS image also plays a crucial issue, which can be addressed by the multi-label classification and Computational Linguistics based image captioning techniques. To achieve this, this paper presents an efficient low complexity encoding technique with multi-label classification and image captioning for UAV based RS images. The presented model primarily involves the low complexity encoder using the Neighborhood Correlation Sequence (NCS) with a burrows wheeler transform (BWT) technique called LCE-BWT for encoding the RS images captured by the UAV. The application of NCS greatly reduces the computation complexity and requires fewer resources for image transmission. Secondly, deep learning (DL) based shallow convolutional neural network for RS image classification (SCNN-RSIC) technique is presented to determine the multiple class labels of the RS image, shows the novelty of the work. Finally, the Computational Linguistics based Bidirectional Encoder Representations from Transformers (BERT) technique is applied for image captioning, to provide a proficient textual description of the RS image. The performance of the presented technique is tested using the UCM dataset. The simulation outcome implied that the presented model has obtained effective compression performance, reconstructed image quality, classification results, and image captioning outcome.

A novel deep learning model for visibility correction of environmental factors in autonomous vehicles

Intelligent vehicles utilize a combination of video-enabled object detection and radar data to traverse safely through surrounding environments. However, since the most momentary missteps in these systems can cause devastating collisions, the margin of error in the software for these systems is small. Furthermore, extenuating weather conditions such as rain, snow, and fog exponentially increase the likelihood of accidents by reducing visibility and increasing the time for detection. In this paper, we hypothesized that a novel object detection system that improves detection accuracy and speed of detection during adverse weather conditions would outperform industry alternatives in an average comparison. To do so, the model employs multiple classical deep learning techniques in two separate sub-modules: a Visibility Correction Module (VCM) and an Object Detection Module (ODM). Firstly, the model employs image classification techniques and masking to identify environmental factors frame-by-frame within an image, and then uses a novel dimensionality reduction network to remove said effects. Next, corrected images are analyzed to classify and label objects within frames. The proposed algorithm achieved an average accuracy of 89.72%, and outperformed industry alternatives in mean accuracy and time for detection, demonstrating the validity and efficiency of utilizing dimensionality reduction to improve object detection.

One-Shot Learning-Based SAR Ship Classification Using New Hybrid Siamese Network

Synthetic aperture radar (SAR) ship classification research mainly focuses on large ships with distinctive features. Also, ship types are expanding nowadays. Traditional standard image classification techniques are not adequate for the effective classification of SAR ships due to their fewer available data in some classes (unbalanced data). We are introducing a one-shot learning-based deep learning model built from scratch to classify 16 ship classes to solve these. It generates embeddings of size 20 and uses distance measure for classification. Also, we propose a preprocessing stage (PP-stage) and feature fusion (FF-stage) technique to improve classification accuracy. Experimental results on the OpenSARShip dataset for ship classification reveal that the accuracy is superior to the other state-of-the-art SAR image classifiers.

Detecting Lung Cancer Using Machine Learning Techniques

In recent days, Internet of Things (IoT) based image classification technique in the healthcare services is becoming a familiar concept that supports the process of detecting cancers with Computer Tomography (CT) images. Lung cancer is one of the perilous diseases that increases the mortality rate exponentially. IoT based image classifiers have the ability to detect cancer at an early stage and increases the life span of a patient. It supports oncologist to monitor and evaluate the health condition of a patient. Also, it can decipher cancer risk marker and act upon them. The process of feature extraction and selection from CT images plays a key role in identifying cancer hot spots. Convolutional Neural Network (CNN) is one of the efficient feature extraction techniques that improves the performance of image classifier by reducing the entropy of image data sets. A Random Forest (RF) classifier is a machine learning technique that can improve its efficiency with the support of CNN. This paper presents an RF classifier with CNN based technique to improve the percentage of accuracy in detecting cancer hot spots with CT images. The experimentation of the proposed approach is based on three dimensions: Feature extraction, selection, and prediction of cancer hot spots. To evaluate the performance of the proposed approach, benchmark image repositories which consists of 3954 images and 50 low dose whole lungs CT scan images are employed. The proposed method achieves an effective result on all test images under different aspects. Consequently, it obtains an average accuracy of 93.25% and an F-measure of 91.75% which is higher than the other methods, comparatively.

Detection of forest fire burn-area using landsat-8 and sentinel-2: a case study of Nivale (Kolhapur) beat of Chandoli National Park, Maharashtra, India

Forests play an important role in maintaining environmental equilibrium in the ecosystem. Forest fires, which can occur for a variety of reasons, are the greatest threat to forests. In order to control forest fires, it is critical to assess the formation and behavioral characteristics of forest fires. Detecting fire areas and the severity is much easier with satellite images obtained with advancing technologies. The main objective of this study is to extract burn area from Landsat-8 and Sentinel-2 satellite images of 2017 using six vegetation indices. Remote sensing data was used to study a forest fire in the Nivale (Kolhapur) beat of Chandoli National Park of Maharashtra. The Department of Divisional Forest Officer (Wildlife)- Chandoli National Park at Karad (DDFO-CNP) provided reference data indicating that the fire had damaged 194 hectares of Scrub forest. In addition, forest fire areas were determined using an objectbased image classification technique. When the findings of the study are compared to the values obtained by DDFO- CNP, it is found that Sentinel-2's object-based analysis provided the highest accuracy with an overall accuracy of 84 % and 0.795 Kappa statistics. Landsat-8 image has 82 % overall accuracy and 0.765 Kappa value. The findings of Sentinel-2 and Landsat-8 spectral indices showed the Sentinel-2 had better results in all indices. Differenced Normalized Difference Vegetation Index (dNDVI) and Relative Difference Normalized Burn Ratio (RdNBR) performed better than other indexes with a difference of only 18.27 and 30.88 hectares respectively. According to the fire severity analysis, a burn area with high intensity in Sentinel-2 was identified as moderate-high in Landsat-8. As per the research findings, sentinel-2 had a high severity area of 28.72 ha and a low severity area of 37.04 ha. It shows that satellite images of Sentinel-2 are highly suitable than Landsat-8 for estimating scrub forest fire areas. Finally, the findings of this study could be useful to forest managers to monitor burn regions quickly after the fire and in reducing severity and frequency of forest fires.

Artificial Intelligence-Based Image Classification Techniques for Hydrologic Applications

ABSTRACT Hydrologic modeling is a complex phenomenon dependent on numerous parameters. Since the estimation of parameters is subjected to high uncertainty due to high spatial variation. Therefore, the accuracy of each parameter becomes prime necessary for hydrologic modeling. Gene Expression Programming (GEP) is employed for the first time for Land Use Land Cover (LULC) classification. In the present study, five AI techniques, namely Support Vector Machine (SVM), Adaptive Neuro-Fuzzy Inference System (ANFIS), the M5 Model tree, the multivariate adaptive regression splines (MARS), and Gene Expression Programming (GEP), were studied comparatively for their image classification capability. Comparison criteria adopted for considered AI techniques were errors estimators’ (omission and commission errors) and accuracy estimators’ (overall accuracy and Kappa coefficient). Based on the obtained results, the performance of the GEP technique is found very much comparable with SVM and ANFIS based on overall & Kappa coefficient (>0.85). GEP has a significant advantage over other techniques in producing mathematical functions for the given set of input and output parameters. The present study recommends the use of the GEP technique for LULC image classification.

Seagrass restoration monitoring and shallow-water benthic habitat mapping through a photogrammetry-based protocol

Seagrasses rank among the most productive yet highly threatened ecosystems on Earth. Loss of seagrass habitat because of anthropogenic disturbances and evidence of their limited resilience have provided the impetus for investigating and monitoring habitat restoration through transplantation programmes. Although Structure from Motion (SfM) photogrammetry is becoming a more and more relevant technique for mapping underwater environments, no standardised methods currently exist to provide 3-dimensional high spatial resolution and accuracy cartographic products for monitoring seagrass transplantation areas. By synthesizing various remote sensing applications, we provide an underwater SfM-based protocol for monitoring large seagrass restoration areas. The data obtained from consumer-grade red-green-blue (RGB) imagery allowed the fine characterization of the seabed by using 3D dense point clouds and raster layers, including orthophoto mosaics and Digital Surface Models (DSM).The integration of high spatial resolution underwater imagery with object-based image classification (OBIA) technique provided a new tool to count transplanted Posidonia oceanica fragments and estimate the bottom coverage expressed as a percentage of seabed covered by such fragments. Finally, the resulting digital maps were integrated into Geographic Information Systems (GIS) to run topographic change detection analysis and evaluate the mean height of transplanted fragments and detect fine-scale changes in seabed vector ruggedness measure (VRM). Our study provides a guide for creating large-scale, replicable and ready-to-use products for a broad range of applications aimed at standardizing monitoring protocols in future seagrass restoration actions.