Search results for: RGB-D images

Authors: Lanchi Xie, Zhihui Li, Zhigang Li, Guiqiang Wang, Lei Xu, Yuwen Yan

Abstract:

Image-based facial features can be classified into category recognition features and individual recognition features. Current automated face recognition systems extract a specific feature vector of different dimensions from a facial image according to their pre-trained neural network. However, to improve the efficiency of parameter calculation, an algorithm generally reduces the image details by pooling. The operation will overlook the details concerned much by forensic experts. In our experiment, we adopted a variety of face recognition algorithms based on deep learning, compared a large number of naturally collected face images with the known data of the same person's frontal ID photos. Downscaling and manual handling were performed on the testing images. The results supported that the facial recognition algorithms based on deep learning detected structural and morphological information and rarely focused on specific markers such as stains and moles. Overall performance, distribution of genuine scores and impostor scores, and likelihood ratios were tested to evaluate the accuracy of biometric systems and forensic experts. Experiments showed that the biometric systems were skilled in distinguishing category features, and forensic experts were better at discovering the individual features of human faces. In the proposed approach, a fusion was performed at the score level. At the specified false accept rate, the framework achieved a lower false reject rate. This paper contributes to improving the interpretability of the objective method of facial comparison and provides a novel method for human-machine collaboration in this field.

Keywords: likelihood ratio, automated facial recognition, facial comparison, biometrics

Procedia PDF Downloads 130

Authors: Annalakshmi G., Sakthivel Murugan S.

Abstract:

This paper presents a local directional encoded derivative binary pattern (LDEDBP) feature extraction method that can be applied for the classification of submarine coral reef images. The classification of coral reef images using texture features is difficult due to the dissimilarities in class samples. In coral reef image classification, texture features are extracted using the proposed method called local directional encoded derivative binary pattern (LDEDBP). The proposed approach extracts the complete structural arrangement of the local region using local binary batten (LBP) and also extracts the edge information using local directional pattern (LDP) from the edge response available in a particular region, thereby achieving extra discriminative feature value. Typically the LDP extracts the edge details in all eight directions. The process of integrating edge responses along with the local binary pattern achieves a more robust texture descriptor than the other descriptors used in texture feature extraction methods. Finally, the proposed technique is applied to an extreme learning machine (ELM) method with a meta-heuristic algorithm known as weighted distance grey wolf optimizer (GWO) to optimize the input weight and biases of single-hidden-layer feed-forward neural networks (SLFN). In the empirical results, ELM-WDGWO demonstrated their better performance in terms of accuracy on all coral datasets, namely RSMAS, EILAT, EILAT2, and MLC, compared with other state-of-the-art algorithms. The proposed method achieves the highest overall classification accuracy of 94% compared to the other state of art methods.

Keywords: feature extraction, local directional pattern, ELM classifier, GWO optimization

Procedia PDF Downloads 163

Authors: Amin Naboureh, Ainong Li, Jinhu Bian, Guangbin Lei, Hamid Ebrahimy

Abstract:

Large-scale land cover mapping has become a new challenge in land change and remote sensing field because of involving a big volume of data. Moreover, selecting the right classification method, especially when there are different types of landscapes in the study area is quite difficult. This paper is an attempt to compare the performance of different machine learning (ML) algorithms for generating a land cover map of the China-Central Asia–West Asia Corridor that is considered as one of the main parts of the Belt and Road Initiative project (BRI). The cloud-based Google Earth Engine (GEE) platform was used for generating a land cover map for the study area from Landsat-8 images (2017) by applying three frequently used ML algorithms including random forest (RF), support vector machine (SVM), and artificial neural network (ANN). The selected ML algorithms (RF, SVM, and ANN) were trained and tested using reference data obtained from MODIS yearly land cover product and very high-resolution satellite images. The finding of the study illustrated that among three frequently used ML algorithms, RF with 91% overall accuracy had the best result in producing a land cover map for the China-Central Asia–West Asia Corridor whereas ANN showed the worst result with 85% overall accuracy. The great performance of the GEE in applying different ML algorithms and handling huge volume of remotely sensed data in the present study showed that it could also help the researchers to generate reliable long-term land cover change maps. The finding of this research has great importance for decision-makers and BRI’s authorities in strategic land use planning.

Keywords: land cover, google earth engine, machine learning, remote sensing

Procedia PDF Downloads 113

Authors: Skyler Kim

Abstract:

An early diagnosis of leukemia has always been a challenge to doctors and hematologists. On a worldwide basis, it was reported that there were approximately 350,000 new cases in 2012, and diagnosing leukemia was time-consuming and inefficient because of an endemic shortage of flow cytometry equipment in current clinical practice. As the number of medical diagnosis tools increased and a large volume of high-quality data was produced, there was an urgent need for more advanced data analysis methods. One of these methods was the AI approach. This approach has become a major trend in recent years, and several research groups have been working on developing these diagnostic models. However, designing and implementing a leukemia diagnostic system in real clinical environments based on a deep learning approach with larger sets remains complex. Leukemia is a major hematological malignancy that results in mortality and morbidity throughout different ages. We decided to select acute lymphocytic leukemia to develop our diagnostic system since acute lymphocytic leukemia is the most common type of leukemia, accounting for 74% of all children diagnosed with leukemia. The results from this development work can be applied to all other types of leukemia. To develop our model, the Kaggle dataset was used, which consists of 15135 total images, 8491 of these are images of abnormal cells, and 5398 images are normal. In this paper, we design and implement a leukemia diagnostic system in a real clinical environment based on deep learning approaches with larger sets. The proposed diagnostic system has the function of detecting and classifying leukemia. Different from other AI approaches, we explore hybrid architectures to improve the current performance. First, we developed two independent convolutional neural network models: VGG19 and ResNet50. Then, using both VGG19 and ResNet50, we developed a hybrid deep learning architecture employing transfer learning techniques to extract features from each input image. In our approach, fusing the features from specific abstraction layers can be deemed as auxiliary features and lead to further improvement of the classification accuracy. In this approach, features extracted from the lower levels are combined into higher dimension feature maps to help improve the discriminative capability of intermediate features and also overcome the problem of network gradient vanishing or exploding. By comparing VGG19 and ResNet50 and the proposed hybrid model, we concluded that the hybrid model had a significant advantage in accuracy. The detailed results of each model’s performance and their pros and cons will be presented in the conference.

Keywords: acute lymphoblastic leukemia, hybrid model, leukemia diagnostic system, machine learning

Procedia PDF Downloads 187

Authors: Fatemeh Parhizkar, Mojtaba Yamani, Abdolla Behboodi, Masoomeh Hashemi

Abstract:

The mangrove forests are natural and valuable gifts that exist in some parts of the world, including Iran. Regarding the threats faced by these forests and the declining area of them all over the world, as well as in Iran, it is very necessary to manage and monitor them. The current study aimed to investigate the changes in mangrove forests and the relationship between these changes and the marine hydrodynamics and coastal morphology in the area between qeshm island and the west coast of the Hormozgan province (i.e. the coastline between Mehran river and Bandar-e Pol port) in the 49-year period. After preprocessing and classifying satellite images using the SVM, MLC, and ANN classifiers and evaluating the accuracy of the maps, the SVM approach with the highest accuracy (the Kappa coefficient of 0.97 and overall accuracy of 98) was selected for preparing the classification map of all images. The results indicate that from 1972 to 1987, the area of these forests have had experienced a declining trend, and in the next years, their expansion was initiated. These forests include the mangrove forests of Khurkhuran wetland, Muriz Deraz Estuary, Haft Baram Estuary, the mangrove forest in the south of the Laft Port, and the mangrove forests between the Tabl Pier, Maleki Village, and Gevarzin Village. The marine hydrodynamic and geomorphological characteristics of the region, such as average intertidal zone, sediment data, the freshwater inlet of Mehran river, wave stability and calmness, topography and slope, as well as mangrove conservation projects make the further expansion of mangrove forests in this area possible. By providing significant and up-to-date information on the development and decline of mangrove forests in different parts of the coast, this study can significantly contribute to taking measures for the conservation and restoration of mangrove forests.

Keywords: mangrove forests, marine hydrodynamics, coastal morphology, west of strait of Hormuz, Iran

Procedia PDF Downloads 95

Authors: Malihe Moeinin, Robert Hill, Ferranti Wong

Abstract:

Background: A white spot lesion (WSL) is defined as subsurface enamel porosity from carious demineralisation on the smooth surfaces of the tooth. It appears as a milky white opacity. Lesions shown an apparently intact surface layer, followed underneath by the more porous lesion body. The small pores within the body of the lesion act as diffusion pathway for both acids and minerals, so allowing the demineralisation of enamel to occur at the advancing front of the lesion. Objectives: The objective is to mapthe porosity and its size on WSL with Focused Ion Bean- Scanning Electron Microscopy (FIB-SEM) Method: The basic method used for FIB-SEM consisted of depositing a one micron thick layer of platinum over 25μmx 25μm of the interest region of enamel. Then, making a rough cut (25μmx 5μmx 20μm) with 3nA current and 30Kv was applied with the help of drift suppression (DS), using a standard “cross-sectional” cutting pattern, which ended at the front of the deposited platinum layer. Two adjacent areas (25μmx 5μmx 20μm) on the both sides of the platinum layer were milled under the same conditions. Subsequent, cleaning cross-sections were applied to polish the sub-surface edge of interest running perpendicular to the surface. The "slice and view" was carried out overnight for milling almost 700 slices with 2Kv and 4nA and taking backscattered (BS) images. Then, images were imported into imageJ and analysed. Results: The prism structure is clearly apparent on FIB-SEM slices of WSL with the dissolution of prism boundaries as well as internal porosity within the prism itself. Porosity scales roughly 100-400nm, which is comparable to the light wavelength (500nm). Conclusion: FIB-SEM is useful to characterize the porosity of WSL and it clearly shows the difference between WSL and normal enamel.

Keywords: white spot lesion, FIB-SEM, enamel porosity, porosity

Procedia PDF Downloads 94

Authors: Jing Li, Xiuli Li

Abstract:

This paper studies a special class of linear codes, called skew cyclic codes, over the ring R= F_q+uF_q+…+u^k-1F_q, where q is a prime power. A Gray map ɸ from R to F_q and a Gray map ɸ' from Rⁿ to Fⁿ_q are defined, as well as an automorphism Θ over R. It is proved that the images of skew cyclic codes over R under map ɸ' and Θ are cyclic codes over F_q, and they still keep the dual relation.

Keywords: skew cyclic code, gray map, automorphism, cyclic code

Procedia PDF Downloads 297

Authors: Martina Depentor

Abstract:

While the cinematic medium contributes to bringing images of a culture to foreign audiences, Audiovisual Translation contributes to deciphering those cultural representations to those same audiences. Through Audiovisual Translation, in fact, elements permeate the reception system and contribute to forging a cultural image of the original/source system in the target/reception system. By analyzing a number of Italian critical reviews, blogs and forum posts, this paper examines the impact and reception in Italy of five of the most successful and influential New Zealand films of the last two decades - An Angel at my Table (1990), The Piano (1993), Heavenly Creatures (1994), Once Were Warriors (1994), Whale Rider (2002) - with the aim of exploring how the adaptation of New Zealand films might condition the representation of New Zealand in the Italian imaginary. The analysis seeks to identify whether a certain degree of cultural loss results from the 'translation' of these films. The films selected share common ground in that they all reveal cultural, social and historical characteristics of New Zealand, from aspects that are unique to this country and that on the surface may render it difficult to penetrate (unfamiliar landscapes, aspects of indigenous culture) to more universal themes (intimate family stories, dysfunctional relationship). They contributed to situating New Zealand on an international stage and to bringing images of the country to many audiences, the Italian one included, with little previous cultural knowledge of the social and political history of New Zealand. Differences in film types pose clearly different levels of interpretative challenges to non-New Zealander audiences, and examples from the films will show how these challenges are or are not overcome if the adaptations display misinterpretations or rendition gaps, and how the process of intercultural transfer further 'domesticates' or 'exoticises' the source culture.

Keywords: audiovisual translation, cultural representation, intercultural transfer, New Zealand Films

Procedia PDF Downloads 301

Authors: P. Karthick, K. Mahesh

Abstract:

Video has become an increasingly significant component of our digital everyday contact. With the advancement of greater contents and shows of the resolution, its significant volume poses serious obstacles to the objective of receiving, distributing, compressing, and revealing video content of high quality. In this paper, we propose the primary beginning to complete a deep video compression model that jointly upgrades all video compression components. The video compression method involves splitting the video into frames, comparing the images using convolutional neural networks (CNN) to remove duplicates, repeating the single image instead of the duplicate images by recognizing and detecting minute changes using generative adversarial network (GAN) and recorded with long short-term memory (LSTM). Instead of the complete image, the small changes generated using GAN are substituted, which helps in frame level compression. Pixel wise comparison is performed using K-nearest neighbours (KNN) over the frame, clustered with K-means, and singular value decomposition (SVD) is applied for each and every frame in the video for all three color channels [Red, Green, Blue] to decrease the dimension of the utility matrix [R, G, B] by extracting its latent factors. Video frames are packed with parameters with the aid of a codec and converted to video format, and the results are compared with the original video. Repeated experiments on several videos with different sizes, duration, frames per second (FPS), and quality results demonstrate a significant resampling rate. On average, the result produced had approximately a 10% deviation in quality and more than 50% in size when compared with the original video.

Keywords: video compression, K-means clustering, convolutional neural network, generative adversarial network, singular value decomposition, pixel visualization, stochastic gradient descent, frame per second extraction, RGB channel extraction, self-detection and deciding system

Procedia PDF Downloads 187

Authors: Ahmed Abdrabou, Medhat Abdalla

Abstract:

The studied stele dates back to Third Intermediate Period (1075-664) B.C in an ancient Egypt. It is made of wood and covered with painted gesso layers. This study aims to use a combination of multi spectral imaging {visible, infrared (IR), Visible-induced infrared luminescence (VIL), Visible-induced ultraviolet luminescence (UVL) and ultraviolet reflected (UVR)}, along with portable x-ray fluorescence in order to map and identify the pigments as well as to provide a deeper understanding of the painting techniques. Moreover; the authors were significantly interested in the identification of wood species. Multispectral imaging acquired in 3 spectral bands, ultraviolet (360-400 nm), visible (400-780 nm) and infrared (780-1100 nm) using (UV Ultraviolet-induced luminescence (UVL), UV Reflected (UVR), Visible (VIS), Visible-induced infrared luminescence (VIL) and Infrared photography. False color images are made by digitally editing the VIS with IR or UV images using Adobe Photoshop. Optical Microscopy (OM), potable X-ray fluorescence spectroscopy (p-XRF) and Fourier Transform Infrared Spectroscopy (FTIR) were used in this study. Mapping and imaging techniques provided useful information about the spatial distribution of pigments, in particular visible-induced luminescence (VIL) which allowed the spatial distribution of Egyptian blue pigment to be mapped and every region containing Egyptian blue, even down to single crystals in some instances, is clearly visible as a bright white area; however complete characterization of the pigments requires the use of p. XRF spectroscopy. Based on the elemental analysis found by P.XRF, we conclude that the artists used mixtures of the basic mineral pigments to achieve a wider palette of hues. Identification of wood species Microscopic identification indicated that the wood used was Sycamore Fig (Ficus sycomorus L.) which is recorded as being native to Egypt and was used to make wooden artifacts since at least the Fifth Dynasty.

Keywords: polychrome wooden stele, multispectral imaging, IR luminescence, Wood identification, Sycamore Fig, p-XRF

Procedia PDF Downloads 264

Authors: Animish Jain

Abstract:

This study deals with using Multiplicative Weight Update within artificial intelligence and machine learning to create models that can diagnose skin cancer using microscopic images of cancer samples. In this study, the multiplicative weight update method is used to take the predictions of multiple models to try and acquire more accurate results. Logistic Regression, Convolutional Neural Network (CNN), and Support Vector Machine Classifier (SVMC) models are employed within the Multiplicative Weight Update system. These models are trained on pictures of skin cancer from the ISIC-Archive, to look for patterns to label unseen scans as either benign or malignant. These models are utilized in a multiplicative weight update algorithm which takes into account the precision and accuracy of each model through each successive guess to apply weights to their guess. These guesses and weights are then analyzed together to try and obtain the correct predictions. The research hypothesis for this study stated that there would be a significant difference in the accuracy of the three models and the Multiplicative Weight Update system. The SVMC model had an accuracy of 77.88%. The CNN model had an accuracy of 85.30%. The Logistic Regression model had an accuracy of 79.09%. Using Multiplicative Weight Update, the algorithm received an accuracy of 72.27%. The final conclusion that was drawn was that there was a significant difference in the accuracy of the three models and the Multiplicative Weight Update system. The conclusion was made that using a CNN model would be the best option for this problem rather than a Multiplicative Weight Update system. This is due to the possibility that Multiplicative Weight Update is not effective in a binary setting where there are only two possible classifications. In a categorical setting with multiple classes and groupings, a Multiplicative Weight Update system might become more proficient as it takes into account the strengths of multiple different models to classify images into multiple categories rather than only two categories, as shown in this study. This experimentation and computer science project can help to create better algorithms and models for the future of artificial intelligence in the medical imaging field.

Keywords: artificial intelligence, machine learning, multiplicative weight update, skin cancer

Procedia PDF Downloads 79

Authors: Li-Kai Kuo, Shyh-Hau Wang

Abstract:

The kinetic trajectory of the median nerve (MN) in the wrist has shown to be capable of being applied to assess the carpal tunnel syndrome (CTS), and was found able to be detected by high-frequency ultrasound image via motion tracking technique. Yet, previous study may not quickly perform the measurement due to the use of a single element transducer for ultrasound image scanning. Therefore, previous system is not appropriate for being applied to clinical application. In the present study, B-mode ultrasound images of the wrist corresponding to movements of fingers from flexion to extension were acquired by clinical applicable real-time scanner. The kinetic trajectories of MN were off-line estimated utilizing two dimensional Baysian speckle tracking (TDBST) technique. The experiments were carried out from ten volunteers by ultrasound scanner at 12 MHz frequency. Results verified from phantom experiments have demonstrated that TDBST technique is able to detect the movement of MN based on signals of the past and present information and then to reduce the computational complications associated with the effect of such image quality as the resolution and contrast variations. Moreover, TDBST technique tended to be more accurate than that of the normalized cross correlation tracking (NCCT) technique used in previous study to detect movements of the MN in the wrist. In response to fingers’ flexion movement, the kinetic trajectory of the MN moved toward the ulnar-palmar direction, and then toward the radial-dorsal direction corresponding to the extensional movement. TDBST technique and the employed ultrasound image scanner have verified to be feasible to sensitively detect the kinetic trajectory and displacement of the MN. It thus could be further applied to diagnose CTS clinically and to improve the measurements to assess 3D trajectory of the MN.

Keywords: baysian speckle tracking, carpal tunnel syndrome, median nerve, motion tracking

Procedia PDF Downloads 495

Authors: Nirjhar Dhang, S. Vinay Kumar

Abstract:

Concrete is a non-homogeneous mix of coarse aggregates, sand, cement, air-voids and interfacial transition zone (ITZ) around aggregates. Adoption of these complex structures and material properties in numerical simulation would lead us to better understanding and design of concrete. In this work, the mesoscale model of concrete has been prepared from X-ray computerized tomography (CT) image. These images are converted into computer model and numerically simulated using commercially available finite element software. The mesoscale models are simulated under the influence of compressive displacement. The effect of shape and distribution of aggregates, continuous and discrete ITZ thickness, voids, and variation of mortar strength has been investigated. The CT scan of concrete cube consists of series of two dimensional slices. Total 49 slices are obtained from a cube of 150mm and the interval of slices comes approximately 3mm. In CT scan images, the same cube can be CT scanned in a non-destructive manner and later the compression test can be carried out in a universal testing machine (UTM) for finding its strength. The image processing and extraction of mortar and aggregates from CT scan slices are performed by programming in Python. The digital colour image consists of red, green and blue (RGB) pixels. The conversion of RGB image to black and white image (BW) is carried out, and identification of mesoscale constituents is made by putting value between 0-255. The pixel matrix is created for modeling of mortar, aggregates, and ITZ. Pixels are normalized to 0-9 scale considering the relative strength. Here, zero is assigned to voids, 4-6 for mortar and 7-9 for aggregates. The value between 1-3 identifies boundary between aggregates and mortar. In the next step, triangular and quadrilateral elements for plane stress and plane strain models are generated depending on option given. Properties of materials, boundary conditions, and analysis scheme are specified in this module. The responses like displacement, stresses, and damages are evaluated by ABAQUS importing the input file. This simulation evaluates compressive strengths of 49 slices of the cube. The model is meshed with more than sixty thousand elements. The effect of shape and distribution of aggregates, inclusion of voids and variation of thickness of ITZ layer with relation to load carrying capacity, stress-strain response and strain localizations of concrete have been studied. The plane strain condition carried more load than plane stress condition due to confinement. The CT scan technique can be used to get slices from concrete cores taken from the actual structure, and the digital image processing can be used for finding the shape and contents of aggregates in concrete. This may be further compared with test results of concrete cores and can be used as an important tool for strength evaluation of concrete.

Keywords: concrete, image processing, plane strain, interfacial transition zone

Procedia PDF Downloads 239

Authors: Joseph Marae Djouda, Ashraf Kasmi, François Hild

Abstract:

Fused filament fabrication is one of the most widespread additive manufacturing techniques because of its low-cost implementation. Its initial development was based on part fabrication with thermoplastic materials. The influence of the manufacturing parameters such as the filament orientation through the nozzle, the deposited layer thickness, or the speed deposition on the mechanical properties of the parts has been widely experimentally investigated. It has been recorded the remarkable variations of the anisotropy in the function of the filament path during the fabrication process. However, there is a lack in the development of constitutive models describing the mechanical properties. In this study, integrated digital image correlation (I-DIC) is used for the identification of mechanical constitutive parameters of two configurations of ABS samples: +/-45° and so-called “oriented deposition.” In this last, the filament was deposited in order to follow the principal strain of the sample. The identification scheme based on the gap reduction between simulation and the experiment directly from images recorded from a single sample (single edge notched tension specimen) is developed. The macroscopic and mesoscopic analysis are conducted from images recorded in both sample surfaces during the tensile test. The elastic and elastoplastic models in isotropic and orthotropic frameworks have been established. It appears that independently of the sample configurations (filament orientation during the fabrication), the elastoplastic isotropic model gives the correct description of the behavior of samples. It is worth noting that in this model, the number of constitutive parameters is limited to the one considered in the elastoplastic orthotropic model. This leads to the fact that the anisotropy of the architectured 3D printed ABS parts can be neglected in the establishment of the macroscopic behavior description.

Keywords: elastic anisotropy, fused filament fabrication, Acrylonitrile butadiene styrene, I-DIC identification

Procedia PDF Downloads 126

Authors: Visvaganthie Moodley

Abstract:

While rhetoric goes back as far as Aristotle who focalised its meaning as the “art of persuasion”, most scholars have focused on elocutio and dispositio canons, neglecting the rhetorical impact of multimodal texts, such as documentaries. Film documentaries are being increasingly rhetoric, often used by wildlife conservationists for influencing people to become more mindful about humanity’s connection with nature. This paper examines the award-winning film documentary, “My Octopus Teacher”, which depicts naturalist, Craig Foster’s unique discovery and relationship with a female octopus in the southern tip of Africa, the Cape of Storms in South Africa. It is anchored in Leech and Short’s (2007) framework of linguistic and stylistic categories – comprising lexical items, grammatical features, figures of speech and other rhetoric features, and cohesiveness – with particular foci on diction, anthropomorphic language, metaphors and symbolism. It also draws on Kress and van Leeuwen’s (2006) multimodal analysis to show how verbal cues (the narrator’s commentary), visual images in motion, visual images as metaphors and symbolism, and aural sensory images such as music and sound synergise for rhetoric effect. In addition, the analysis of “My Octopus Teacher” is guided by Nichol’s (2010) narrative theory; features of a documentary which foregrounds the credibility of the narrative as a text that represents real events with real people; and its modes of construction, viz., the poetic mode, the expository mode, observational mode and participatory mode, and their integration – forging documentaries as multimodal texts. This paper presents a multimodal rhetoric discussion on the sequence of salient episodes captured in the slow moving one-and-a-half-hour documentary. These are: (i) The prologue: on the brink of something extraordinary; (ii) The day it all started; (iii) The narrator’s turmoil: getting back into the ocean; (iv) The incredible encounter with the octopus; (v) Establishing a relationship; (vi) Outwitting the predatory pyjama shark; (vii) The cycle of life; and (viii) The conclusion: lessons from an octopus. The paper argues that wildlife documentaries, characterized by plausibility and which provide researchers the lens to examine the ideologies about animals and humans, offer an assimilation of the various senses – vocal, visual and audial – for engaging viewers in stylized compelling way; they have the ability to persuade people to think and act in particular ways. As multimodal texts, with its use of lexical items; diction; anthropomorphic language; linguistic, visual and aural metaphors and symbolism; and depictions of anthropocentrism, wildlife documentaries are powerful resources for promoting wildlife conservation and conscientizing people of the need for establishing a harmonious relationship with nature and humans alike.

Keywords: documentaries, multimodality, rhetoric, style, wildlife, conservation

Procedia PDF Downloads 94

Authors: Waqqas-ur-Rehman Butt, Martin Servin, Marion Pause

Abstract:

In recent years, object detection has gained much attention and very encouraging research area in the field of computer vision. The robust object boundaries detection in an image is demanded in numerous applications of human computer interaction and automated surveillance systems. Many methods and approaches have been developed for automatic object detection in various fields, such as automotive, quality control management and environmental services. Inappropriately, to the best of our knowledge, object detection under illumination with shadow consideration has not been well solved yet. Furthermore, this problem is also one of the major hurdles to keeping an object detection method from the practical applications. This paper presents an approach to automatic object detection in images under non-standardized environmental conditions. A key challenge is how to detect the object, particularly under uneven illumination conditions. Image capturing conditions the algorithms need to consider a variety of possible environmental factors as the colour information, lightening and shadows varies from image to image. Existing methods mostly failed to produce the appropriate result due to variation in colour information, lightening effects, threshold specifications, histogram dependencies and colour ranges. To overcome these limitations we propose an object detection algorithm, with pre-processing methods, to reduce the interference caused by shadow and illumination effects without fixed parameters. We use the Y CrCb colour model without any specific colour ranges and predefined threshold values. The segmented object regions are further classified using morphological operations (Erosion and Dilation) and contours. Proposed approach applied on a large image data set acquired under various environmental conditions for wood stack detection. Experiments show the promising result of the proposed approach in comparison with existing methods.

Keywords: image processing, illumination equalization, shadow filtering, object detection

Procedia PDF Downloads 216

Authors: Seyed Mehdi Seyed Alizadeh, Christoph Arns, Shane Latham

Abstract:

Geomechanical characterization of rocks in detail and its possible implications on flow properties is an important aspect of reservoir characterization workflow. In order to gain more understanding of the microstructure evolution of reservoir rocks under stress a series of axisymmetric triaxial tests were performed on two different analogue rock samples. In-situ compression tests were coupled with high resolution micro-Computed Tomography to elucidate the changes in the pore/grain network of the rocks under pressurized conditions. Two outcrop sandstones were chosen in the current study representing a various cementation status of well-consolidated and weakly-consolidated granular system respectively. High resolution images were acquired while the rocks deformed in a purpose-built compression cell. A detailed analysis of the 3D images in each series of step-wise compression tests (up to the failure point) was conducted which includes the registration of the deformed specimen images with the reference pristine dry rock image. Digital Image Correlation (DIC) technique based on the intensity of the registered 3D subsets and particle tracking are utilized to map the displacement fields in each sample. The results suggest the complex architecture of the localized shear zone in well-cemented Bentheimer sandstone whereas for the weakly-consolidated Castlegate sandstone no discernible shear band could be observed even after macroscopic failure. Post-mortem imaging a sister plug from the friable rock upon undergoing continuous compression reveals signs of a shear band pattern. This suggests that for friable sandstones at small scales loading mode may affect the pattern of deformation. Prior to mechanical failure, the continuum digital image correlation approach can reasonably capture the kinematics of deformation. As failure occurs, however, discrete image correlation (i.e. particle tracking) reveals superiority in both tracking the grains as well as quantifying their kinematics (in terms of translations/rotations) with respect to any stage of compaction. An attempt was made to quantify the displacement field in compression using continuum Digital Image Correlation which is based on the reference and secondary image intensity correlation. Such approach has only been previously applied to unconsolidated granular systems under pressure. We are applying this technique to sandstones with various degrees of consolidation. Such element of novelty will set the results of this study apart from previous attempts to characterize the deformation pattern in consolidated sands.

Keywords: deformation mechanism, displacement field, shear behavior, triaxial compression, X-ray micro-CT

Procedia PDF Downloads 189

Authors: Mahdieh Farzin Asanjan, Erkan Unal Mumcuoglu

Abstract:

Mitochondria are cytoplasmic organelles of the cell, which have a significant role in the variety of cellular metabolic functions. Mitochondria act as the power plants of the cell and are surrounded by two membranes. Significant morphological alterations are often due to changes in mitochondrial functions. A powerful technique in order to study the three-dimensional (3D) structure of mitochondria and its alterations in disease states is Electron microscope tomography. Detection of mitochondria in electron microscopy images due to the presence of various subcellular structures and imaging artifacts is a challenging problem. Another challenge is that each image typically contains more than one mitochondrion. Hand segmentation of mitochondria is tedious and time-consuming and also special knowledge about the mitochondria is needed. Fully automatic segmentation methods lead to over-segmentation and mitochondria are not segmented properly. Therefore, semi-automatic segmentation methods with minimum manual effort are required to edit the results of fully automatic segmentation methods. Here two editing tools were implemented by applying spline surface dragging and interactive live-wire segmentation tools. These editing tools were applied separately to the results of fully automatic segmentation. 3D extension of these tools was also studied and tested. Dice coefficients of 2D and 3D for surface dragging using splines were 0.93 and 0.92. This metric for 2D and 3D for live-wire method were 0.94 and 0.91 respectively. The root mean square symmetric surface distance values of 2D and 3D for surface dragging was measured as 0.69, 0.93. The same metrics for live-wire tool were 0.60 and 2.11. Comparing the results of these editing tools with the results of automatic segmentation method, it shows that these editing tools, led to better results and these results were more similar to ground truth image but the required time was higher than hand-segmentation time

Keywords: medical image segmentation, semi-automatic methods, transmission electron microscopy, surface dragging using splines, live-wire

Procedia PDF Downloads 169

Authors: Abderrahim Elmoataz

Abstract:

More and more contemporary applications involve data in the form of functions defined on irregular and topologically complicated domains (images, meshs, points clouds, networks, etc). Such data are not organized as familiar digital signals and images sampled on regular lattices. However, they can be conveniently represented as graphs where each vertex represents measured data and each edge represents a relationship (connectivity or certain affinities or interaction) between two vertices. Processing and analyzing these types of data is a major challenge for both image and machine learning communities. Hence, it is very important to transfer to graphs and networks many of the mathematical tools which were initially developed on usual Euclidean spaces and proven to be efficient for many inverse problems and applications dealing with usual image and signal domains. Historically, the main tools for the study of graphs or networks come from combinatorial and graph theory. In recent years there has been an increasing interest in the investigation of one of the major mathematical tools for signal and image analysis, which are Partial Differential Equations (PDEs) variational methods on graphs. The normalized p-laplacian operator has been recently introduced to model a stochastic game called tug-of-war-game with noise. Part interest of this class of operators arises from the fact that it includes, as particular case, the infinity Laplacian, the mean curvature operator and the traditionnal Laplacian operators which was extensiveley used to models and to solve problems in image processing. The purpose of this paper is to introduce and to study a new class of normalized p-Laplacian on graphs. The introduction is based on the extension of p-harmonious function introduced in as discrete approximation for both infinity Laplacian and p-Laplacian equations. Finally, we propose to use these operators as a framework for solving many inverse problems in image processing.

Keywords: normalized p-laplacian, image processing, stochastic game, inverse problems

Procedia PDF Downloads 512

Authors: El Asnaoui Khalid, Aksasse Brahim, Ouanan Mohammed

Abstract:

This paper presents an unsupervised color image segmentation method. It is based on a hierarchical analysis of 2-D histogram in RGB color space. This histogram minimizes storage space of images and thus facilitates the operations between them. The improved segmentation approach shows a better identification of objects in a color image and, at the same time, the system is fast.

Keywords: image segmentation, hierarchical analysis, 2-D histogram, classification

Procedia PDF Downloads 380

Authors: Maria A. Castillo H., Andres R. Leandro, Jose F. Bienvenido B.

Abstract:

During the last 20 years, the globalization of economies, population growth, and the increase in the consumption of fresh agricultural products have generated greater demand for ornamentals, flowers, fresh fruits, and vegetables, mainly from tropical areas. This market situation has demanded greater competitiveness and control over production, with more efficient protected agriculture technologies, which provide greater productivity and allow us to guarantee the quality and quantity that is required in a constant and sustainable way. Guatemala, located in the north of Central America, is one of the largest exporters of agricultural products in the region and exports fresh vegetables, flowers, fruits, ornamental plants, and foliage, most of which were grown in greenhouses. Although there are no official agricultural statistics on greenhouse production, several thesis works, and congress reports have presented consistent estimates. A wide range of protection structures and roofing materials are used, from the most basic and simple ones for rain control to highly technical and automated structures connected with remote sensors for monitoring and control of crops. With this breadth of technological models, it is necessary to analyze georeferenced data related to the cultivated area, to the different existing models, and to the covering materials, integrated with altitude, climate, and soil data. The georeferenced registration of the production units, the data collection with digital tools, the use of satellite images, and geographic information systems (GIS) provide reliable tools to elaborate more complete, agile, and dynamic information maps. This study details a methodology proposed for gathering georeferenced data of high protection structures (greenhouses) in Guatemala, structured in four phases: diagnosis of available information, the definition of the geographic frame, selection of satellite images, and integration with an information system geographic (GIS). It especially takes account of the actual lack of complete data in order to obtain a reliable decision-making system; this gap is solved through the proposed methodology. A summary of the results is presented in each phase, and finally, an evaluation with some improvements and tentative recommendations for further research is added. The main contribution of this study is to propose a methodology that allows to reduce the gap of georeferenced data in protected agriculture in this specific area where data is not generally available and to provide data of better quality, traceability, accuracy, and certainty for the strategic agricultural decision öaking, applicable to other crops, production models and similar/neighboring geographic areas.

Keywords: greenhouses, protected agriculture, GIS, Guatemala, satellite image, digital tools, precision agriculture

Procedia PDF Downloads 194

Authors: Sylwia Orzechowska, Andrzej Wróbel, Eugeniusz Rokita

Abstract:

The mineralization is a curious problem of connective tissues. Factors which may play a decisive role in the regulation of the yellow ligaments (YL) mineralization are still open questions. The aim of the studies was a detailed description of the chemical composition and morphology of mineral deposits in the human yellow ligaments. Investigations of the structural features of deposits were used to explain the impact of various factors on mineralization process. The studies were carried out on 24 YL samples, surgically removed from patients suffer from spinal canal stenosis and the patients who sustained a trauma. The micro-computed tomography was used to describe the morphology of mineral deposits. The X-ray fluorescence method and Fourier transform infrared spectroscopy were applied to determine the chemical composition of the samples. In order to eliminate the effect of blur in microtomographic images, the correction method of partial volume effect was used. The mineral deposits appear in 60% of YL samples, both in patients with a stenosis and following injury. The mineral deposits have a heterogeneous structure and they are a mixture of the tissue and mineral grains. The volume of mineral grains amounts to (1.9 ± 3.4)*10-3 mm3 while the density distribution of grains occurs in two distinct ranges (1.75 - 2.15 and 2.15-2.5) g/cm3. Application of the partial volume effect correction allows accurate calculations by eliminating the averaging effect of gray levels in tomographic images. The B-type carbonate-containing hydroxyapatite constitutes the mineral phase of majority YLs. The main phase of two samples was calcium pyrophosphate dihydrate (CPPD). The elemental composition of minerals in all samples is almost identical. This pathology may be independent on the spine diseases and it does not evoke canal stenosis. The two ranges of grains density indicate two stages of grains growth and the degree of maturity. The presence of CPPD crystals may coexist with other pathologies.

Keywords: FTIR, micro-tomography, mineralization, spinal ligaments

Procedia PDF Downloads 377

Authors: Paola Utreras, Yazmina Olmos, Loreto Sanhueza

Abstract:

This work has the purpose of study and incorporate Problem Based Learning (PBL) for engineering students, through the analysis of several thermal images of dwellings located in different geographical points of the Region de los Ríos, Chile. The students analyze how heat is transferred in and out of the houses and how is the relation between heat transfer and climatic conditions that affect each zone. As a result of this activity students are able to acquire significant learning in the unit of heat and temperature, and manage to reverse previous conceptual errors related with energy, temperature and heat. In addition, student are able to generate prototype solutions to increase thermal efficiency using low cost materials. Students make public their results in a report using scientific writing standards and in a science fair open to the entire university community. The methodology used to measure previous Conceptual Errors has been applying diagnostic tests with everyday questions that involve concepts of heat, temperature, work and energy, before the unit. After the unit the same evaluation is done in order that themselves are able to evidence the evolution in the construction of knowledge. As a result, we found that in the initial test, 90% of the students showed deficiencies in the concepts previously mentioned, and in the subsequent test 47% showed deficiencies, these percent ages differ between students who carry out the course for the first time and those who have performed this course previously in a traditional way. The methodology used to measure Significant Learning has been by comparing results in subsequent courses of thermodynamics among students who have received problem based learning and those who have received traditional training. We have observe that learning becomes meaningful when applied to the daily lives of students promoting internalization of knowledge and understanding through critical thinking.

Keywords: engineering students, heat flow, problem-based learning, thermal images

Procedia PDF Downloads 231

Authors: Yonggu Jang, Jisong Ryu, Woosik Lee

Abstract:

The study aims to address the limitations of existing underground facility exploration equipment in terms of exploration depth range, relative depth measurement, data processing time, and human-centered ground penetrating radar image interpretation. The study proposed the use of 3D absolute positioning technology to develop a precision underground facility exploration system. The aim of this study is to establish a precise exploration system for underground facilities based on 3D absolute positioning technology, which can accurately survey up to a depth of 5m and measure the 3D absolute location of precise underground facilities. The study developed software and hardware technologies to build the precision exploration system. The software technologies developed include absolute positioning technology, ground surface location synchronization technology of GPR exploration equipment, GPR exploration image AI interpretation technology, and integrated underground space map-based composite data processing technology. The hardware systems developed include a vehicle-type exploration system and a cart-type exploration system. The data was collected using the developed exploration system, which employs 3D absolute positioning technology. The GPR exploration images were analyzed using AI technology, and the three-dimensional location information of the explored precise underground facilities was compared to the integrated underground space map. The study successfully developed a precision underground facility exploration system based on 3D absolute positioning technology. The developed exploration system can accurately survey up to a depth of 5m and measure the 3D absolute location of precise underground facilities. The system comprises software technologies that build a 3D precise DEM, synchronize the GPR sensor's ground surface 3D location coordinates, automatically analyze and detect underground facility information in GPR exploration images and improve accuracy through comparative analysis of the three-dimensional location information, and hardware systems, including a vehicle-type exploration system and a cart-type exploration system. The study's findings and technological advancements are essential for underground safety management in Korea. The proposed precision exploration system significantly contributes to establishing precise location information of underground facility information, which is crucial for underground safety management and improves the accuracy and efficiency of exploration. The study addressed the limitations of existing equipment in exploring underground facilities, proposed 3D absolute positioning technology-based precision exploration system, developed software and hardware systems for the exploration system, and contributed to underground safety management by providing precise location information. The developed precision underground facility exploration system based on 3D absolute positioning technology has the potential to provide accurate and efficient exploration of underground facilities up to a depth of 5m. The system's technological advancements contribute to the establishment of precise location information of underground facility information, which is essential for underground safety management in Korea.

Keywords: 3D absolute positioning, AI interpretation of GPR exploration images, complex data processing, integrated underground space maps, precision exploration system for underground facilities

Procedia PDF Downloads 62

Authors: Adel Khorramrouz, Monireh Ahmadi Bani, Ehsan Norouzi, Morvarid Lalenoor

Abstract:

Background: Plantar pressure distribution measurement has been used for a long time to assess foot disorders. Plantar pressure is an important component affecting the foot and ankle function and Changes in plantar pressure distribution could indicate various foot and ankle disorders. Morphologic and mechanical properties of the foot may be important factors affecting the plantar pressure distribution. Accurate and early measurement may help to reduce the prevalence of pes planus. With recent developments in technology, new techniques such as machine learning have been used to assist clinicians in predicting patients with foot disorders. Significance of the study: This study proposes a neural network learning-based flat foot classification methodology using static foot pressure distribution. Methodologies: Data were collected from 895 patients who were referred to a foot clinic due to foot disorders. Patients with pes planus were labeled by an experienced physician based on clinical examination. Then all subjects (with and without pes planus) were evaluated for static plantar pressures distribution. Patients who were diagnosed with the flat foot in both feet were included in the study. In the next step, the leg length was normalized and the network was trained for plantar pressure mapping images. Findings: From a total of 895 image data, 581 were labeled as pes planus. A computational neural network (CNN) ran to evaluate the performance of the proposed model. The prediction accuracy of the basic CNN-based model was performed and the prediction model was derived through the proposed methodology. In the basic CNN model, the training accuracy was 79.14%, and the test accuracy was 72.09%. Conclusion: This model can be easily and simply used by patients with pes planus and doctors to predict the classification of pes planus and prescreen for possible musculoskeletal disorders related to this condition. However, more models need to be considered and compared for higher accuracy.

Keywords: foot disorder, machine learning, neural network, pes planus

Procedia PDF Downloads 360

Authors: Balakrishna Shetty

Abstract:

Introduction: Stem Cell Imaging is a challenging field since the advent of Stem Cell treatment in humans. Series of research on tagging and tracking the stem cells has not been very effective. The present study is an effort by the authors to track the stem cells injected into calf muscles by Magnetic Resonance Diffusion Weighted Imaging. Materials and methods: Stem Cell injection deep into the calf muscles of patients with peripheral vascular disease is one of the recent treatment modalities followed in our institution. 5 patients who underwent deep intramuscular injection of stem cells as treatment were included for this study. Pre and two hours Post injection MRI of bilateral calf regions was done using 1.5 T Philips Achieva, 16 channel system using 16 channel torso coils. Axial STIR, Axial Diffusion weighted images with b=0 and b=1000 values with back ground suppression (DWIBS sequence of Philips MR Imaging Systems) were obtained at 5 mm interval covering the entire calf. The invert images were obtained for better visualization. 120ml of autologous bone marrow derived stem cells were processed and enriched under c-GMP conditions and reduced to 40ml solution containing mixture of above stem cells. Approximately 40 to 50 injections, each containing 0.75ml of processed stem cells, was injected with marked grids over the calf region. Around 40 injections, each of 1ml normal saline, is injected into contralateral leg as control. Results: Significant Diffusion hyper intensity is noted at the site of injected stem cells. No hyper intensity noted before the injection and also in the control side where saline was injected conclusion: This is one of the earliest studies in literature showing diffusion hyper intensity in intramuscularly injected stem cells. The advantages and deficiencies in this study will be discussed during the presentation.

Keywords: stem cells, imaging, DWI, peripheral vascular disease

Procedia PDF Downloads 74

Authors: Shahla Bukhtair

Abstract:

A paradigm presents a consensus view of a particular or collective community, accepted into by the members of that community, either consciously pronounced or, more likely, simply assumed and not intentionally acknowledged but is articulated. Paradigm shift is based on the behavioral attitude of the community. Change is inexorable. The world is suffering with the innovative creation of globalization. Media boosted this paradigm shift all over the world. Globalization is a vigorous process which impacts differentially on various cultures around the world. The outcome of the globalization is permeates cultural boundaries and in the process results in the spread of Western ideologies and values across the world. The term flourished in 20th century. Globalization is regarded as having substantial impact on such crises through its encouragement of conflicts rather than conciliation; through opportunities of expression, various groups get benefit with it. Identity crisis refers to inflexible mechanism i.e. cultural and political conflicts among polarized groups, which struggle with each other over the definition of a national identity. Violence is not only a kind of physical but it also psychological as well. Due to identity crisis, a person is having an issue of fear, anxiety, and lack of security. Everything has negative and positive aspects. Newspaper columns, magazine articles, films, made-for-TV movies, television special reports, and talk shows are all public arenas where images of political agenda of their own interest are constructed, debated, and reproduced. From these resources, individuals construct their own conceptions of what is normal and acceptable. This bias affects images in the media, and in turn has a negative effect on public development in a society. This paper investigates the relationship between globalization and cultural war, identity crisis and the role of violence. Objectives: - To determine which type of media plays an important role in shaping perceptions and attitudes of public negatively; - To analyze the impact of globalization on identity crisis, violence and global culture (positive and negative).

Keywords: paradigm shift, globalization, identity crisis, cultural war

Procedia PDF Downloads 366

Authors: E. Farg, M. M. El-Sharkawy, M. S. Mostafa, S. M. Arafat

Abstract:

Mapping water stress provides important baseline data for sustainable agriculture. Recent developments in the new Sentinel-1 data which allow the acquisition of high resolution images and varied polarization capabilities. This study was conducted to detect and quantify vegetation water content from canopy backscatter for extracting spatial information to encourage drought mapping activities throughout new reclaimed sandy soils in western Nile delta, Egypt. The performance of radar imagery in agriculture strongly depends on the sensor polarization capability. The dual mode capabilities of Sentinel-1 improve the ability to detect water stress and the backscatter from the structure components improves the identification and separation of vegetation types with various canopy structures from other features. The fieldwork data allowed identifying of water stress zones based on land cover structure; those classes were used for producing harmonious water stress map. The used analysis techniques and results show high capability of active sensors data in water stress mapping and monitoring especially when integrated with multi-spectral medium resolution images. Also sub soil drip irrigation systems cropped areas have lower drought and water stress than center pivot sprinkler irrigation systems. That refers to high level of evaporation from soil surface in initial growth stages. Results show that high relationship between vegetation indices such as Normalized Difference Vegetation Index NDVI the observed radar backscattering. In addition to observational evidence showed that the radar backscatter is highly sensitive to vegetation water stress, and essentially potential to monitor and detect vegetative cover drought.

Keywords: canopy backscatter, drought, polarization, NDVI

Procedia PDF Downloads 144

Authors: Susana Barreto

Abstract:

This paper explores a pedagogical exercise that employs family photos as catalysts for teaching visual analysis and inspiring academic writing among MA students. The study aimed to achieve two primary objectives: to impart students with the skills of analyzing images or artifacts and to ignite their writing for research purposes. Conducted at Viana Polytechnic in Portugal, the exercise involved two classes on Arts Management and Art Education Master course comprising approximately twenty students from diverse academic backgrounds, including Economics, Design, Fine Arts, and Sociology, among others. The exploratory exercise involved selecting an old family photo, analyzing its content and context, and deconstructing the chosen images in an intuitive and systematic manner. Students were encouraged to engage in photo elicitation, seeking insights from family/friends to gain multigenerational perspectives on the images. The feedback received from this exercise was consistently positive, largely due to the personal connection students felt with the objects of analysis. Family photos, with their emotional significance, fostered deeper engagement and motivation in the learning process. Furthermore, visual analysing family photos stimulated critical thinking as students interpreted the composition, subject matter, and potential meanings embedded in the images. This practice enhanced their ability to comprehend complex visual representations and construct compelling visual narratives, thereby facilitating the writing process. The exercise also facilitated the identification of patterns, similarities, and differences by comparing different family photos, leading to a more comprehensive analysis of visual elements and themes. Throughout the exercise, students found analyzing their own photographs both enjoyable and insightful. They progressed through preliminary analysis, explored content and context, and artfully interwove these components. Additionally, students experimented with various techniques such as converting photos to black and white, altering framing angles, and adjusting sizes to unveil hidden meanings.The methodology employed included observation, documental analysis of written reports, and student interviews. By including students from diverse academic backgrounds, the study enhanced its external validity, enabling a broader range of perspectives and insights during the exercise. Furthermore, encouraging students to seek multigenerational perspectives from family and friends added depth to the analysis, enriching the learning experience and broadening the understanding of the cultural and historical context associated with the family photos Highlighting the emotional significance of these family photos and the personal connection students felt with the objects of analysis fosters a deeper connection to the subject matter. Moreover, the emphasis on stimulating critical thinking through the analysis of composition, subject matter, and potential meanings in family photos suggests a targeted approach to developing analytical skills. This improvement focuses specifically on critical thinking and visual analysis, enhancing the overall quality of the exercise. Additionally, the inclusion of a step where students compare different family photos to identify patterns, similarities, and differences further enhances the depth of the analysis. This comparative approach adds a layer of complexity to the exercise, ultimately leading to a more comprehensive understanding of visual elements and themes. The expected results of this study will culminate in a set of practical recommendations for implementing this exercise in academic settings.

Keywords: visual analysis, academic writing, pedagogical exercise, family photos

Procedia PDF Downloads 59

Authors: Khairani Binti Supyan, Fatimah Khalid, Mas Rina Mustaffa, Azreen Bin Azman, Amirul Azuani Romle

Abstract:

Perennial plant classification plays a significant role in various agricultural and environmental applications, assisting in plant identification, disease detection, and biodiversity monitoring. Nevertheless, attaining high accuracy in perennial plant image classification remains challenging due to the complex variations in plant appearance, the diverse range of environmental conditions under which images are captured, and the inherent variability in image quality stemming from various factors such as lighting conditions, camera settings, and focus. This paper proposes an adaptation approach to optimize perennial plant image classification by fine-tuning the pre-trained DNNs model. This paper explores the efficacy of fine-tuning prevalent architectures, namely VGG16, ResNet50, and InceptionV3, leveraging transfer learning to tailor the models to the specific characteristics of perennial plant datasets. A subset of the MYLPHerbs dataset consisted of 6 perennial plant species of 13481 images under various environmental conditions that were used in the experiments. Different strategies for fine-tuning, including adjusting learning rates, training set sizes, data augmentation, and architectural modifications, were investigated. The experimental outcomes underscore the effectiveness of fine-tuning deep neural networks for perennial plant image classification, with ResNet50 showcasing the highest accuracy of 99.78%. Despite ResNet50's superior performance, both VGG16 and InceptionV3 achieved commendable accuracy of 99.67% and 99.37%, respectively. The overall outcomes reaffirm the robustness of the fine-tuning approach across different deep neural network architectures, offering insights into strategies for optimizing model performance in the domain of perennial plant image classification.

Keywords: perennial plants, image classification, deep neural networks, fine-tuning, transfer learning, VGG16, ResNet50, InceptionV3

Procedia PDF Downloads 64