Search results for: photographic images

Authors: Jennifer Buz, Alvin Spivey

Abstract:

The most common greenhouse gas emitted through human activities, carbon dioxide (CO2), is naturally consumed by plants during photosynthesis. This process is actively being monetized by companies wishing to offset their carbon dioxide emissions. For example, companies are now able to purchase protections for vegetated land due-to-be clear cut or purchase barren land for reforestation. Therefore, by actively preventing the destruction/decay of plant matter or by introducing more plant matter (reforestation), a company can theoretically offset some of their emissions. One of the biggest issues in the carbon credit market is validating and verifying carbon offsets. There is a need for a system that can accurately and frequently ensure that the areas sold for carbon credits have the vegetation mass (and therefore for carbon offset capability) they claim. Traditional techniques for measuring vegetation mass and determining health are costly and require many person-hours. Orbital Sidekick offers an alternative approach that accurately quantifies carbon mass and assesses vegetation health through satellite hyperspectral imagery, a technique which enables us to remotely identify material composition (including plant species) and condition (e.g., health and growth stage). How much carbon a plant is capable of storing ultimately is tied to many factors, including material density (primarily species-dependent), plant size, and health (trees that are actively decaying are not effectively storing carbon). All of these factors are capable of being observed through satellite hyperspectral imagery. This abstract focuses on speciation. To build a species classification model, we matched pixels in our remote sensing imagery to plants on the ground for which we know the species. To accomplish this, we collaborated with the researchers at the Teakettle Experimental Forest. Our remote sensing data comes from our airborne “Kato” sensor, which flew over the study area and acquired hyperspectral imagery (400-2500 nm, 472 bands) at ~0.5 m/pixel resolution. Coverage of the entire teakettle experimental forest required capturing dozens of individual hyperspectral images. In order to combine these images into a mosaic, we accounted for potential variations of atmospheric conditions throughout the data collection. To do this, we ran an open source atmospheric correction routine called ISOFIT1 (Imaging Spectrometer Optiman FITting), which converted all of our remote sensing data from radiance to reflectance. A database of reflectance spectra for each of the tree species within the study area was acquired using the Teakettle stem map and the geo-referenced hyperspectral images. We found that a wide variety of machine learning classifiers were able to identify the species within our images with high (>95%) accuracy. For the most robust quantification of carbon mass and the best assessment of the health of a vegetated area, speciation is critical. Through the use of high resolution hyperspectral data, ground-truth databases, and complex analytical techniques, we are able to determine the species present within a pixel to a high degree of accuracy. These species identifications will feed directly into our carbon mass model.

Keywords: hyperspectral, satellite, carbon, imagery, python, machine learning, speciation

Procedia PDF Downloads 107

Authors: Tahir Ali Akbar, Hirra Jabbar

Abstract:

Land cover change analysis plays a significant role in understanding the trends of urban sprawl and land use transformation due to anthropogenic activities. In this study, the spatio-temporal dynamics of major land covers were analyzed in the last twenty years (1988-2016) for District Lahore located in the Punjab Province of Pakistan. The Landsat satellite imageries were downloaded from USGS Global Visualization Viewer of Earth Resources Observation and Science Center located in Sioux Falls, South Dakota USA. The imageries included: (i) Landsat TM-5 for 1988 and 2001; and (ii) Landsat-8 OLI for 2016. The raw digital numbers of Landsat-5 images were converted into spectral radiance and then planetary reflectance. The digital numbers of Landsat-8 image were directly converted into planetary reflectance. The normalized difference vegetation index (NDVI) was used to classify the processed images into six major classes of water, buit-up, barren land, shrub and grassland, sparse vegetation and dense vegetation. The NDVI output results were improved by visual interpretation using high-resolution satellite imageries. The results indicated that the built-up areas were increased to 21% in 2016 from 10% in 1988. The decrease in % areas was found in case of water, barren land and shrub & grassland. There were improvements in percentage of areas for the vegetation. The increasing trend of urban sprawl for Lahore requires implementation of GIS based spatial planning, monitoring and management system for its sustainable development.

Keywords: land cover changes, NDVI, remote sensing, urban sprawl

Procedia PDF Downloads 305

Authors: Alessandro Preziosi, Antonio Prioletti, Luca Castangia

Abstract:

Deformable part models achieve high precision in pedestrian recognition, but all publicly available implementations are too slow for real-time applications. We implemented a deformable part model algorithm fast enough for real-time use by exploiting information about the camera position and orientation. This implementation is both faster and more precise than alternative DPM implementations. These results are obtained by computing convolutions in the frequency domain and using lookup tables to speed up feature computation. This approach is almost an order of magnitude faster than the reference DPM implementation, with no loss in precision. Knowing the position of the camera with respect to horizon it is also possible prune many hypotheses based on their size and location. The range of acceptable sizes and positions is set by looking at the statistical distribution of bounding boxes in labelled images. With this approach it is not needed to compute the entire feature pyramid: for example higher resolution features are only needed near the horizon. This results in an increase in mean average precision of 5% and an increase in speed by a factor of two. Furthermore, to reduce misdetections involving small pedestrians near the horizon, input images are supersampled near the horizon. Supersampling the image at 1.5 times the original scale, results in an increase in precision of about 4%. The implementation was tested against the public KITTI dataset, obtaining an 8% improvement in mean average precision over the best performing DPM-based method. By allowing for a small loss in precision computational time can be easily brought down to our target of 100ms per image, reaching a solution that is faster and still more precise than all publicly available DPM implementations.

Keywords: autonomous vehicles, deformable part model, dpm, pedestrian detection, real time

Procedia PDF Downloads 267

Authors: U. Pruthviraj, Praveen Kumar R. A. K. Athul, K. V. Gangadharan, S. Rao Shrikantha

Abstract:

An open source based autonomous unmanned marine surface vehicle (UMSV) is developed for some of the marine applications such as pollution control, environmental monitoring and thermal imaging. A double rotomoulded hull boat is deployed which is rugged, tough, quick to deploy and moves faster. It is suitable for environmental monitoring, and it is designed for easy maintenance. A 2HP electric outboard marine motor is used which is powered by a lithium-ion battery and can also be charged from a solar charger. All connections are completely waterproof to IP67 ratings. In full throttle speed, the marine motor is capable of up to 7 kmph. The motor is integrated with an open source based controller using cortex M4F for adjusting the direction of the motor. This UMSV can be operated by three modes: semi-autonomous, manual and fully automated. One of the channels of a 2.4GHz radio link 8 channel transmitter is used for toggling between different modes of the USMV. In this electric outboard marine motor an on board GPS system has been fitted to find the range and GPS positioning. The entire system can be assembled in the field in less than 10 minutes. A Flir Lepton thermal camera core, is integrated with a 64-bit quad-core Linux based open source processor, facilitating real-time capturing of thermal images and the results are stored in a micro SD card which is a data storage device for the system. The thermal camera is interfaced to an open source processor through SPI protocol. These thermal images are used for finding oil spills and to look for people who are drowning at low visibility during the night time. A Real Time clock (RTC) module is attached with the battery to provide the date and time of thermal images captured. For the live video feed, a 900MHz long range video transmitter and receiver is setup by which from a higher power output a longer range of 40miles has been achieved. A Multi-parameter probe is used to measure the following parameters: conductivity, salinity, resistivity, density, dissolved oxygen content, ORP (Oxidation-Reduction Potential), pH level, temperature, water level and pressure (absolute).The maximum pressure it can withstand 160 psi, up to 100m. This work represents a field demonstration of an open source based autonomous navigation system for a marine surface vehicle.

Keywords: open source, autonomous navigation, environmental monitoring, UMSV, outboard motor, multi-parameter probe

Procedia PDF Downloads 226

Authors: Yufeng Li, Yinan Ma, Jing Wu, Chengnian Long

Abstract:

Infrared object detection aims at separating small and dim target from clutter background and its capabilities extend beyond the limits of visible light, making it invaluable in a wide range of applications such as improving safety, security, efficiency, and functionality. However, existing methods are usually sensitive to the noise of the input infrared image, leading to a decrease in target detection accuracy and an increase in the false alarm rate in high-noise environments. To address this issue, an infrared small target detection algorithm called YOLO-IR is proposed in this paper to improve the robustness to high infrared noise. To address the problem that high noise significantly reduces the clarity and reliability of target features in infrared images, we design a soft-threshold coordinate attention mechanism to improve the model’s ability to extract target features and its robustness to noise. Since the noise may overwhelm the local details of the target, resulting in the loss of small target features during depth down-sampling, we propose a deep and shallow feature fusion neck to improve the detection accuracy. In addition, because the generalized Intersection over Union (IoU)-based loss functions may be sensitive to noise and lead to unstable training in high-noise environments, we introduce a Wasserstein-distance based loss function to improve the training of the model. The experimental results show that YOLO-IR achieves a 5.0% improvement in recall and a 6.6% improvement in F1-score over existing state-of-art model.

Keywords: infrared small target detection, high noise, robustness, soft-threshold coordinate attention, feature fusion

Procedia PDF Downloads 45

Authors: Ethan James

Abstract:

Over one billion people worldwide suffer from some level of vision loss or blindness as a result of progressive retinal diseases. Many patients, particularly in developing areas, are incorrectly diagnosed or undiagnosed whatsoever due to unconventional diagnostic tools and screening methods. Artificial intelligence (AI) based on deep learning (DL) convolutional neural networks (CNN) have recently gained a high interest in ophthalmology for its computer-imaging diagnosis, disease prognosis, and risk assessment. Optical coherence tomography (OCT) is a popular imaging technique used to capture high-resolution cross-sections of retinas. In ophthalmology, DL has been applied to fundus photographs, optical coherence tomography, and visual fields, achieving robust classification performance in the detection of various retinal diseases including macular degeneration, diabetic retinopathy, and retinitis pigmentosa. However, there is no complete diagnostic model to analyze these retinal images that provide a diagnostic accuracy above 90%. Thus, the purpose of this project was to develop an AI model that utilizes machine learning techniques to automatically diagnose specific retinal diseases from OCT scans. The algorithm consists of neural network architecture that was trained from a dataset of over 20,000 real-world OCT images to train the robust model to utilize residual neural networks with cyclic pooling. This DL model can ultimately aid ophthalmologists in diagnosing patients with these retinal diseases more quickly and more accurately, therefore facilitating earlier treatment, which results in improved post-treatment outcomes.

Keywords: artificial intelligence, deep learning, imaging, medical devices, ophthalmic devices, ophthalmology, retina

Procedia PDF Downloads 163

Authors: Yolanda Havenga

Abstract:

Being a competent nurse requires clinical, general, and moral competencies. Moral competence is a culmination of moral perceptions, moral judgment, moral behaviour, and moral identity. Moral identity is the values, images, and fundamental principles held in the collective minds and memories of nurses about what it means to be a ‘good nurse’. It is important to explore and describe South African nurses’ moral identities and excavate the post-colonial counter-narrative to nurses moral identities as a better understanding of these identities will enable means to positively address nurses’ moral behaviours. This study explored the moral identity of nurses within the South African context. A qualitative approach was followed triangulating with phenomenological and narrative designs with the same purposively sampled group of professional nurses. In-depth interviews were conducted until saturation of data occurred about the sampled nurses lived experiences of being a nurse in South Africa. They were probed about their core personal-, social-, and professional values. Data were analysed based on the steps used by Colaizzi. These nurses were then asked to write a narrative telling a personal story that portrayed a significant time in their professional career that defines their identity as a nurse. This data were analysed using a critical narrative approach and findings of the two sets of data were merged. Ethical approval was obtained and approval from all relevant gate keepers. In the findings, themes emerged related to personal, social and professional values, images and fundamental principles of being a nurse within the South African context. The findings of this study will inform a future national study including a representative sample of South African nurses.

Keywords: moral behaviour, moral identity, nurses, qualitative research

Procedia PDF Downloads 270

Authors: Zhongmin Wang, Wudong Fan, Hengshan Zhang, Yimin Zhou

Abstract:

In data-driven prognostic methods, the prediction accuracy of the estimation for remaining useful life of bearings mainly depends on the performance of health indicators, which are usually fused some statistical features extracted from vibrating signals. However, the existing health indicators have the following two drawbacks: (1) The differnet ranges of the statistical features have the different contributions to construct the health indicators, the expert knowledge is required to extract the features. (2) When convolutional neural networks are utilized to tackle time-frequency features of signals, the time-series of signals are not considered. To overcome these drawbacks, in this study, the method combining convolutional neural network with gated recurrent unit is proposed to extract the time-frequency image features. The extracted features are utilized to construct health indicator and predict remaining useful life of bearings. First, original signals are converted into time-frequency images by using continuous wavelet transform so as to form the original feature sets. Second, with convolutional and pooling layers of convolutional neural networks, the most sensitive features of time-frequency images are selected from the original feature sets. Finally, these selected features are fed into the gated recurrent unit to construct the health indicator. The results state that the proposed method shows the enhance performance than the related studies which have used the same bearing dataset provided by PRONOSTIA.

Keywords: continuous wavelet transform, convolution neural net-work, gated recurrent unit, health indicators, remaining useful life

Procedia PDF Downloads 118

Authors: Silvija Šiljeg, Ivan Marić, Ante Šiljeg

Abstract:

The accessibility of urban green spaces (UGS) is an integral element in the quality of life. Due to rapid urbanization, UGS studies have become a key element in urban planning. The potential benefits of space for its inhabitants are frequently analysed. A functional transport network system and the optimal spatial distribution of urban green surfaces are the prerequisites for maintaining the environmental equilibrium of the urban landscape. An accessibility analysis was conducted as part of the Urban Green Belts Project (UGB). The development of a GIS database for Zadar was the first step in generating the UGS accessibility indicator. Data were collected using the supervised classification method of multispectral LANDSAT images and manual vectorization of digital orthophoto images (DOF). An analysis of UGS accessibility according to the ANGst standard was conducted in the first phase of research. The accessibility indicator was generated on the basis of seven objective measurements, which included average UGS surface per capita and accessibility according to six functional levels of green surfaces. The generated indicator was compared with subjective measurements obtained by conducting a survey (718 respondents) within statistical units. The collected data reflected individual assessments and subjective evaluations of UGS accessibility. This study highlighted the importance of using objective and subjective measures in the process of understanding the accessibility of urban green surfaces. It may be concluded that when evaluating UGS accessibility, residents emphasize the immediate residential environment, ignoring higher UGS functional levels. It was also concluded that large areas of UGS within a city do not necessarily generate similar satisfaction with accessibility. The heterogeneity of output results may serve as guidelines for the further development of a functional UGS city network.

Keywords: urban green spaces (UGS), accessibility indicator, subjective and objective measurements, Zadar

Procedia PDF Downloads 240

Authors: Keerti Kembhavi, M. R. Archana, V. Anjaneyappa

Abstract:

Most asphalt pavement condition evaluation use rating frameworks in which asphalt pavement distress is estimated by type, extent, and severity. Rating is carried out by the pavement condition rating (PCR), which is tedious and expensive. This paper presents the development of a low-cost technique for image pavement distress analysis that permits the identification of pothole and cracks. The paper explores the application of image processing tools for the detection of potholes and cracks. Longitudinal cracking and pothole are detected using Fuzzy-C- Means (FCM) and proceeded with the Spectral Theory algorithm. The framework comprises three phases, including image acquisition, processing, and extraction of features. A digital camera (Gopro) with the holder is used to capture pavement distress images on a moving vehicle. FCM classifier and Spectral Theory algorithms are used to compute features and classify the longitudinal cracking and pothole. The Matlab2016Ra Image preparing tool kit utilizes performance analysis to identify the viability of pavement distress on selected urban stretches of Bengaluru city, India. The outcomes of image evaluation with the utilization semi-computerized image handling framework represented the features of longitudinal crack and pothole with an accuracy of about 80%. Further, the detected images are validated with the actual dimensions, and it is seen that dimension variability is about 0.46. The linear regression model y=1.171x-0.155 is obtained using the existing and experimental / image processing area. The R2 correlation square obtained from the best fit line is 0.807, which is considered in the linear regression model to be ‘large positive linear association’.

Keywords: crack detection, pothole detection, spectral clustering, fuzzy-c-means

Procedia PDF Downloads 168

Authors: Bensaid A., Mostephaoui T., Nedjai R.

Abstract:

Nowadays, wetlands are the subject of contradictory debates opposing scientific, political and administrative meanings. Indeed, given their multiple services (drinking water, irrigation, hydrological regulation, mineral, plant and animal resources...), wetlands concentrate many socio-economic and biodiversity issues. In some regions, they can cover vast areas (>100 thousand ha) of the landscape, such as the Camargue area in the south of France, inside the Rhone delta. The high biological productivity of wetlands, the strong natural selection pressures and the diversity of aquatic environments have produced many species of plants and animals that are found nowhere else. These environments are tremendous carbon sinks and biodiversity reserves depending on their age, composition and surrounding environmental conditions, wetlands play an important role in global climate projections. Covering more than 3% of the earth's surface, wetlands have experienced since the beginning of the 1990s a tremendous revival of interest, which has resulted in the multiplication of inventories, scientific studies and management experiments. The geographical and physical characteristics of the wetlands of the central region conceal a large number of natural habitats that harbour a great biological diversity. These wetlands, one of the natural habitats, are still influenced by human activities, especially agriculture, which affects its layout and functioning. In this perspective, decision-makers need to delimit spatial objects (natural habitats) in a certain way to be able to take action. Thus, wetlands are no exception to this rule even if it seems to be a difficult exercise to delimit a type of environment as whose main characteristic is often to occupy the transition between aquatic and terrestrial environment. However, it is possible to map wetlands with databases, derived from the interpretation of photos and satellite images, such as the European database Corine Land cover, which allows quantifying and characterizing for each place the characteristic wetland types. Scientific studies have shown limitations when using high spatial resolution images (SPOT, Landsat, ASTER) for the identification and characterization of small wetlands (1 hectare). To address this limitation, it is important to note that these wetlands generally represent spatially complex features. Indeed, the use of very high spatial resolution images (>3m) is necessary to map small and large areas. However, with the recent evolution of artificial intelligence (AI) and deep learning methods for satellite image processing have shown a much better performance compared to traditional processing based only on pixel structures. Our research work is also based on spectral and textural analysis on THR images (Spot and IRC orthoimage) using two object-oriented approaches, the nearest neighbour approach (k-NN) and the Super Vector Machine approach (SVM). The k-NN approach gave good results for the delineation of wetlands (wet marshes and moors, ponds, artificial wetlands water body edges, ponds, mountain wetlands, river edges and brackish marshes) with a kappa index higher than 85%.

Keywords: land development, GIS, sand dunes, segmentation, remote sensing

Procedia PDF Downloads 46

Authors: Ochuko K. Overen, Edson L. Meyer

Abstract:

Uncontrolled heat transfer between the inner and outer space of low-cost housings through the thermal envelope result in indoor thermal discomfort. As a result, an excessive amount of energy is consumed for space heating and cooling. Thermo-optical properties are the ability of paints to reduce the rate of heat transfer through the thermal envelope. The aim of this study is to analyze the thermal performance of a low-cost house with its walls inner surface coated with transparent infrared reflective paint. The thermo-optical properties of the paint were analyzed using Scanning Electron Microscopy/ Energy Dispersive X-ray spectroscopy (SEM/EDX), Fourier Transform Infra-Red (FTIR) and thermal photographic technique. Meteorological indoor and ambient parameters such as; air temperature, relative humidity, solar radiation, wind speed and direction of a low-cost house in Golf-course settlement, South Africa were monitored. The monitoring period covers both winter and summer period before and after coating. The thermal performance of the coated walls was evaluated using time lag and decrement factor. The SEM image shows that the coat is transparent to light. The presence of Al as Al2O and other elements were revealed by the EDX spectrum. Before coating, the average decrement factor of the walls in summer was found to be 0.773 with a corresponding time lag of 1.3 hours. In winter, the average decrement factor and corresponding time lag were 0.467 and 1.6 hours, respectively. After coating, the average decrement factor and corresponding time lag were 0.533 and 2.3 hour, respectively in summer. In winter, an average decrement factor of 1.120 and corresponding time lag of 3 hours was observed. The findings show that the performance of the coats is influenced by the seasons. With a 74% reduction in decrement factor and 1.4 time lag increase in winter, it implies that the coatings have more ability to retain heat within the inner space of the house than preventing heat flow into the house. In conclusion, the results have shown that transparent infrared reflective paint has the ability to reduce the propagation of heat flux through building walls. Hence, it can serve as a remedy to the poor thermal performance of low-cost housings in South Africa.

Keywords: energy efficiency, decrement factor, low-cost housing, paints, rural development, thermal comfort, time lag

Procedia PDF Downloads 272

Authors: Gurukiran Veerapur, Nytik Birudavolu, Seetharam U. N., Chandravva Hebbi, R. Praneeth Reddy

Abstract:

This work presents a robust approach for the recognition of characters in Telugu and Kannada, two South Indian scripts with structural similarities in characters. To recognize the characters exhaustive datasets are required, but there are only a few publicly available datasets. As a result, we decided to create a dataset for one language (source language),train the model with it, and then test it with the target language.Telugu is the target language in this work, whereas Kannada is the source language. The suggested method makes use of Canny edge features to increase character identification accuracy on pictures with noise and different lighting. A dataset of 45,150 images containing printed Kannada characters was created. The Nudi software was used to automatically generate printed Kannada characters with different writing styles and variations. Manual labelling was employed to ensure the accuracy of the character labels. The deep learning models like CNN (Convolutional Neural Network) and Visual Attention neural network (VAN) are used to experiment with the dataset. A Visual Attention neural network (VAN) architecture was adopted, incorporating additional channels for Canny edge features as the results obtained were good with this approach. The model's accuracy on the combined Telugu and Kannada test dataset was an outstanding 97.3%. Performance was better with Canny edge characteristics applied than with a model that solely used the original grayscale images. The accuracy of the model was found to be 80.11% for Telugu characters and 98.01% for Kannada words when it was tested with these languages. This model, which makes use of cutting-edge machine learning techniques, shows excellent accuracy when identifying and categorizing characters from these scripts.

Keywords: base characters, modifiers, guninthalu, aksharas, vattakshara, VAN

Procedia PDF Downloads 38

Authors: Abdelrahman Elsehsah

Abstract:

The main objective of this study is to assess changes in the water capacity of Aswan High Dam Lake (AHDL) caused by evaporation and seepage losses. To achieve this objective, a comprehensive methodology was employed. The methodology involves acquiring Sentinel-3 imagery and extracting the surface area of the lake using remote sensing techniques. Using water areas calculated from sentinel images, collected field data, and the lake’s water balance equation, monthly evaporation and seepage losses were estimated for the years 2021 and 2022. Based on the water balance method results, the average monthly evaporation losses for the year 2021 were estimated to be around 1.41 billion cubic meters (Bm3), which closely matches the estimates provided by the Ministry of Water Resources and Irrigation (MWRI) annual reports (approximately 1.37 Bm3 in the same year). This means that the water balance method slightly overestimated the monthly evaporation losses by about 2.92%. Similarly, the average monthly seepage losses for the year 2022 were estimated to be around 0.005 Bm3, while the MWRI reports indicated approximately 0.0046 Bm3. By another means, the water balance method overestimated the monthly seepage losses by about 8.70%. Furthermore, the study found that the average monthly evaporation rate within AHDL was 210.88 mm/month, which closely aligns with the computed value of approximately 204.9 mm/month by AHDA. These findings indicated that the applied water balance method, utilizing remote sensing and field data, is a reliable tool for estimating monthly evaporation and seepage losses as well as monthly evaporation rates in AHDL.

Keywords: Aswan high dam lake, remote sensing, water balance equation, seepage loss, evaporation loss

Procedia PDF Downloads 10

Authors: James Gallagher, Phillip Benachour

Abstract:

As location aware mobile technologies become ever more omnipresent, the prospect of exploiting their context awareness to enforce learning approaches thrives. Utilizing the growing acceptance of ubiquitous computing, and the steady progress both in accuracy and battery usage of pervasive devices, we present a case study of remote location viewing, how the application can be utilized to support mobile learning in situ using an existing scenario. Through the case study we introduce a new innovative application: Mobipeek based around a request/response protocol for the viewing of a remote location and explore how this can apply both as part of a teacher lead activity and informal learning situations. The system developed allows a user to select a point on a map, and send a request. Users can attach messages alongside time and distance constraints. Users within the bounds of the request can respond with an image, and accompanying message, providing context to the response. This application can be used alongside a structured learning activity such as the use of mobile phone cameras outdoors as part of an interactive lesson. An example of a learning activity would be to collect photos in the wild about plants, vegetation, and foliage as part of a geography or environmental science lesson. Another example could be to take photos of architectural buildings and monuments as part of an architecture course. These images can be uploaded then displayed back in the classroom for students to share their experiences and compare their findings with their peers. This can help to fosters students’ active participation while helping students to understand lessons in a more interesting and effective way. Mobipeek could augment the student learning experience by providing further interaction with other peers in a remote location. The activity can be part of a wider study between schools in different areas of the country enabling the sharing and interaction between more participants. Remote location viewing can be used to access images in a specific location. The choice of location will depend on the activity and lesson. For example architectural buildings of a specific period can be shared between two or more cities. The augmentation of the learning experience can be manifested in the different contextual and cultural influences as well as the sharing of images from different locations. In addition to the implementation of Mobipeek, we strive to analyse this application, and a subset of other possible and further solutions targeted towards making learning more engaging. Consideration is given to the benefits of such a system, privacy concerns, and feasibility of widespread usage. We also propose elements of “gamification”, in an attempt to further the engagement derived from such a tool and encourage usage. We conclude by identifying limitations, both from a technical, and a mobile learning perspective.

Keywords: context aware, location aware, mobile learning, remote viewing

Procedia PDF Downloads 278

Authors: Bryce R. Jolley, Michael Uchic

Abstract:

This study presents initial results toward the correlative characterization of inherent defects of Ti-6Al-4V additive manufacture (AM). X-Ray Computed Tomography (CT) defect data are compared and correlated with microscopic photographs obtained via automated serial sectioning. The metal AM specimen was manufactured out of Ti-6Al-4V virgin powder to specified dimensions. A post-contour was applied during the fabrication process with a speed of 1050 mm/s, power of 260 W, and a width of 140 µm. The specimen was stress relief heat-treated at 16°F for 3 hours. Microfocus CT imaging was accomplished on the specimen within a predetermined region of the build. Microfocus CT imaging was conducted with parameters optimized for Ti-6Al-4V additive manufacture. After CT imaging, a modified RoboMet. 3D version 2 was employed for serial sectioning and optical microscopy characterization of the same predetermined region. Automated montage capture with sub-micron resolution, bright-field reflection, 12-bit monochrome optical images were performed in an automated fashion. These optical images were post-processed to produce 2D and 3D data sets. This processing included thresholding and segmentation to improve visualization of defect features. The defects observed from optical imaging were compared and correlated with the defects observed from CT imaging over the same predetermined region of the specimen. Quantitative results of area fraction and equivalent pore diameters obtained via each method are presented for this correlation. It is shown that Microfocus CT imaging does not capture all inherent defects within this Ti-6Al-4V AM sample. Best practices for this correlative effort are also presented as well as the future direction of research resultant from this current study.

Keywords: additive manufacture, automated serial sectioning, computed tomography, nondestructive evaluation

Procedia PDF Downloads 131

Authors: Ati Moncef

Abstract:

We describe in this paper the results of a phantom of dynamics renal with MAG3. Our phantom consisted of (tow shaped of kidneys, 1 liver). These phantoms were scanned with static and dynamic protocols and compared with clinical data. in a normal conditions we use our phantoms it's possible to acquire a renal images when we can be compared with clinical scintigraphy. In conclusion, Renal phantom also can use in the quality control of a renal scintigraphy.

Keywords: Renal scintigraphy, MAG3, Nuclear medicine, Gamma Camera.

Procedia PDF Downloads 391

Authors: Halimatu S. Abdullahi, Ray E. Sheriff, Fatima Mahieddine

Abstract:

Agriculture is essential to the continuous existence of human life as they directly depend on it for the production of food. The exponential rise in population calls for a rapid increase in food with the application of technology to reduce the laborious work and maximize production. Technology can aid/improve agriculture in several ways through pre-planning and post-harvest by the use of computer vision technology through image processing to determine the soil nutrient composition, right amount, right time, right place application of farm input resources like fertilizers, herbicides, water, weed detection, early detection of pest and diseases etc. This is precision agriculture which is thought to be solution required to achieve our goals. There has been significant improvement in the area of image processing and data processing which has being a major challenge. A database of images is collected through remote sensing, analyzed and a model is developed to determine the right treatment plans for different crop types and different regions. Features of images from vegetations need to be extracted, classified, segmented and finally fed into the model. Different techniques have been applied to the processes from the use of neural network, support vector machine, fuzzy logic approach and recently, the most effective approach generating excellent results using the deep learning approach of convolution neural network for image classifications. Deep Convolution neural network is used to determine soil nutrients required in a plantation for maximum production. The experimental results on the developed model yielded results with an average accuracy of 99.58%.

Keywords: convolution, feature extraction, image analysis, validation, precision agriculture

Procedia PDF Downloads 303

Authors: Levent Basayigit, Sinan Demir, Burhan Kara, Yusuf Ucar

Abstract:

Cannabis is mostly used for drug production. In some countries, an excessive amount of illegal cannabis is cultivated and sold. Most of the illegal cannabis cultivation occurs on the lands far from settlements. In farmlands, it is cultivated with other crops. In this method, cannabis is surrounded by tall plants like corn and sunflower. It is also cultivated with tall crops as the mixed culture. The common method of the determination of the illegal cultivation areas is to investigate the information obtained from people. This method is not sufficient for the determination of illegal cultivation in remote areas. For this reason, more effective methods are needed for the determination of illegal cultivation. Remote Sensing is one of the most important technologies to monitor the plant growth on the land. The aim of this study is to monitor cannabis cultivation area using satellite imagery. The main purpose of this study was to develop an applicable method for monitoring the cannabis cultivation. For this purpose, cannabis was grown as single or surrounded by the corn and sunflower in plots. The morphological characteristics of cannabis were recorded two times per month during the vegetation period. The spectral signature library was created with the spectroradiometer. The parcels were monitored with high-resolution satellite imagery. With the processing of satellite imagery, the cultivation areas of cannabis were classified. To separate the Cannabis plots from the other plants, the multiresolution segmentation algorithm was found to be the most successful for classification. WorldView Improved Vegetative Index (WV-VI) classification was the most accurate method for monitoring the plant density. As a result, an object-based classification method and vegetation indices were sufficient for monitoring the cannabis cultivation in multi-temporal Earthwiev images.

Keywords: Cannabis, drug, remote sensing, object-based classification

Procedia PDF Downloads 262

Authors: Lela Gadrani, Mariam Tsitsagi

Abstract:

Human is a part of the urban landscape and responsible for it. Urbanization of cities includes the longest phase; thus none of the environment ever undergoes such anthropogenic impact as the area of large cities. The post-Soviet period is very interesting in terms of scientific research. The changes that have occurred in the cities since the collapse of the Soviet Union have not yet been analyzed best to our knowledge. In this context, the aim of this paper is to analyze the changes in the land use of the three large cities of Georgia (Tbilisi, Kutaisi, Batumi). Tbilisi as a capital city, Batumi as a port city, and Kutaisi as a former industrial center. Data used during the research process are conventionally divided into satellite and supporting materials. For this purpose, the largest topographic maps (1:10 000) of all three cities were analyzed, Tbilisi General Plans (1896, 1924), Tbilisi and Kutaisi historical maps. The main emphasis was placed on the classification of Landsat images. In this case, we have classified the images LULC (LandUse / LandCover) of all three cities taken in 1987 and 2016 using the supervised and unsupervised methods. All the procedures were performed in the programs: Arc GIS 10.3.1 and ENVI 5.0. In each classification we have singled out the following classes: built-up area, water bodies, agricultural lands, green cover and bare soil, and calculated the areas occupied by them. In order to check the validity of the obtained results, additionally we used the higher resolution images of CORONA and Sentinel. Ultimately we identified the changes that took place in the land use in the post-Soviet period in the above cities. According to the results, a large wave of changes touched Tbilisi and Batumi, though in different periods. It turned out that in the case of Tbilisi, the area of developed territory has increased by 13.9% compared to the 1987 data, which is certainly happening at the expense of agricultural land and green cover, in particular, the area of agricultural lands has decreased by 4.97%; and the green cover by 5.67%. It should be noted that Batumi has obviously overtaken the country's capital in terms of development. With the unaided eye it is clear that in comparison with other regions of Georgia, everything is different in Batumi. In fact, Batumi is an unofficial summer capital of Georgia. Undoubtedly, Batumi’s development is very important both in economic and social terms. However, there is a danger that in the uneven conditions of urban development, we will eventually get a developed center - Batumi, and multiple underdeveloped peripheries around it. Analysis of the changes in the land use is of utmost importance not only for quantitative evaluation of the changes already implemented, but for future modeling and prognosis of urban development. Raster data containing the classes of land use is an integral part of the city's prognostic models.

Keywords: analysis, geo information system, remote sensing, LULC

Procedia PDF Downloads 441

Authors: Natascha C. D'Amico, Enzo Grossi, Giovanni Valbusa, Ala Malasevschi, Gianpiero Cardone, Sergio Papa

Abstract:

Purpose: To characterize, through a radiomic approach, the nature of areas classified PI-RADS (Prostate Imaging Reporting and Data System) 3/5, recognized in multiparametric prostate magnetic resonance with T2-weighted (T2w), diffusion and perfusion sequences with paramagnetic contrast. Methods and Materials: 24 cases undergoing multiparametric prostate MR and biopsy were admitted to this pilot study. Clinical outcome of the PI-RADS 3/5 was found through biopsy, finding 8 malignant tumours. The analysed images were acquired with a Philips achieva 1.5T machine with a CE- T2-weighted sequence in the axial plane. Semi-automatic tumour segmentation was carried out on MR images using 3DSlicer image analysis software. 45 shape-based, intensity-based and texture-based features were extracted and represented the input for preprocessing. An evolutionary algorithm (a TWIST system based on KNN algorithm) was used to subdivide the dataset into training and testing set and select features yielding the maximal amount of information. After this pre-processing 20 input variables were selected and different machine learning systems were used to develop a predictive model based on a training testing crossover procedure. Results: The best machine learning system (three-layers feed-forward neural network) obtained a global accuracy of 90% ( 80 % sensitivity and 100% specificity ) with a ROC of 0.82. Conclusion: Machine learning systems coupled with radiomics show a promising potential in distinguishing benign from malign tumours in PI-RADS 3/5 areas.

Keywords: machine learning, MR prostate, PI-Rads 3, radiomics

Procedia PDF Downloads 176

Authors: Antonio Monteiro, Jorge Azevedo, Severiano Silva, Alfredo Teixeira

Abstract:

The objective of this study was to compare the carcass and in vivo real-time ultrasound measurements (RTU) and their capacity to predict the composition of Serrana goats up to 40% of maturity. Twenty one females (11.1 ± 3.97 kg) and Twenty one males (15.6 ± 5.38 kg) were utilized to made in vivo measurements with a 5 MHz probe (ALOKA 500V scanner) at the 9th-10th, 10th-11th thoracic vertebrae (uT910 and uT1011, respectively), at the 1st- 2nd, 3rd-4th, and 4th-5th lumbar vertebrae (uL12, ul34 and uL45, respectively) and also at the 3rd-4th sternebrae (EEST). It was recorded the images of RTU measurements of Longissimus thoracis et lumborum muscle (LTL) depth (EM), width (LM), perimeter (PM), area (AM) and subcutaneous fat thickness (SFD) above the LTL, as well as the depth of tissues of the sternum (EEST) between the 3rd-4th sternebrae. All RTU images were analyzed using the ImageJ software. After slaughter, the carcasses were stored at 4 ºC for 24 h. After this period the carcasses were divided and the left half was entirely dissected into muscle, dissected fat (subcutaneous fat plus intermuscular fat) and bone. Prior to the dissection measurements equivalent to those obtained in vivo with RTU were recorded. Using the Statistica 5, correlation and regression analyses were performed. The prediction of carcass composition was achieved by stepwise regression procedure, with live weight and RTU measurements with and without transformation of variables to the same dimension. The RTU and carcass measurements, except for SFD measurements, showed high correlation (r > 0.60, P < 0.001). The RTU measurements and the live weight, showed ability to predict carcass composition on muscle (R2 = 0.99, P < 0.001), subcutaneous fat (R2 = 0.41, P < 0.001), intermuscular fat (R2 = 0.84, P < 0.001), dissected fat (R2 = 0.71, P < 0.001) and bone (R2 = 0.94, P < 0.001). The transformation of variables allowed a slight increase of precision, but with the increase in the number of variables, with the exception of subcutaneous fat prediction. In vivo measurements by RTU can be applied to predict kid goat carcass composition, from 5 measurements of RTU and the live weight.

Keywords: carcass, goats, real time, ultrasound

Procedia PDF Downloads 248

Authors: Cristina Diaz, Lucia Perez-Gandarillas, Gonzalo Garcia-Fuentes, Simone Visigalli, Roberto Canziani, Giuseppe Di Florio, Paolo Gronchi

Abstract:

During their lifetime, industrial machines are often subjected to chemical, mechanical and thermal extreme conditions. In some cases, the loss of efficiency comes from the degradation of the surface as a result of its exposition to abrasive environments that can cause wear. This is a common problem to be solved in industries of diverse nature such as food, paper or concrete industries, among others. For this reason, a good selection of the material is of high importance. In the machine design context, stainless steels such as AISI 304 and 316 are widely used. However, the severity of the external conditions can require additional protection for the steel and sometimes coating solutions are demanded in order to extend the lifespan of these materials. Therefore, the development of effective coatings with high wear resistance is of utmost technological relevance. In this research, bilayer coatings made of Titanium-Tantalum, Titanium-Niobium, Titanium-Hafnium, and Titanium-Zirconium have been developed using magnetron sputtering configuration by PVD (Physical Vapor Deposition) technology. Their tribological behavior has been measured and evaluated under different environmental conditions. Two kinds of steels were used as substrates: AISI 304, AISI 316. For the comparison with these materials, titanium alloy substrate was also employed. Regarding the characterization, wear rate and friction coefficient were evaluated by a tribo-tester, using a pin-on-ball configuration with different lubricants such as tomato sauce, wine, olive oil, wet compost, a mix of sand and concrete with water and NaCl to approximate the results to real extreme conditions. In addition, topographical images of the wear tracks were obtained in order to get more insight of the wear behavior and scanning electron microscope (SEM) images were taken to evaluate the adhesion and quality of the coating. The characterization was completed with the measurement of nanoindentation hardness and elastic modulus. Concerning the results, thicknesses of the samples varied from 100 nm (Ti-Zr layer) to 1.4 µm (Ti-Hf layer) and SEM images confirmed that the addition of the Ti layer improved the adhesion of the coatings. Moreover, results have pointed out that these coatings have increased the wear resistance in comparison with the original substrates under environments of different severity. Furthermore, nanoindentation hardness results showed an improvement of the elastic strain to failure and a high modulus of elasticity (approximately 200 GPa). As a conclusion, Ti-Ta, Ti-Zr, Ti-Nb, and Ti-Hf are very promising and effective coatings in terms of tribological behavior, improving considerably the wear resistance and friction coefficient of typically used machine materials.

Keywords: coating, stainless steel, tribology, wear

Procedia PDF Downloads 141

Authors: Ali Ashtiani, Hamid Shirazi

Abstract:

This paper reviews the status of existing practice and research related to measuring pavement cracking and using crack density as a pavement surface evaluation protocol. Crack density for pavement evaluation is currently not widely used within the airport community and its use by the highway community is limited. However, surface cracking is a distress that is closely monitored by airport staff and significantly influences the development of maintenance, rehabilitation and reconstruction plans for airport pavements. Therefore crack density has the potential to become an important indicator of pavement condition if the type, severity and extent of surface cracking can be accurately measured. A pavement distress survey is an essential component of any pavement assessment. Manual crack surveying has been widely used for decades to measure pavement performance. However, the accuracy and precision of manual surveys can vary depending upon the surveyor and performing surveys may disrupt normal operations. Given the variability of manual surveys, this method has shown inconsistencies in distress classification and measurement. This can potentially impact the planning for pavement maintenance, rehabilitation and reconstruction and the associated funding strategies. A substantial effort has been devoted for the past 20 years to reduce the human intervention and the error associated with it by moving toward automated distress collection methods. The automated methods refer to the systems that identify, classify and quantify pavement distresses through processes that require no or very minimal human intervention. This principally involves the use of a digital recognition software to analyze and characterize pavement distresses. The lack of established protocols for measurement and classification of pavement cracks captured using digital images is a challenge to developing a reliable automated system for distress assessment. Variations in types and severity of distresses, different pavement surface textures and colors and presence of pavement joints and edges all complicate automated image processing and crack measurement and classification. This paper summarizes the commercially available systems and technologies for automated pavement distress evaluation. A comprehensive automated pavement distress survey involves collection, interpretation, and processing of the surface images to identify the type, quantity and severity of the surface distresses. The outputs can be used to quantitatively calculate the crack density. The systems for automated distress survey using digital images reviewed in this paper can assist the airport industry in the development of a pavement evaluation protocol based on crack density. Analysis of automated distress survey data can lead to a crack density index. This index can be used as a means of assessing pavement condition and to predict pavement performance. This can be used by airport owners to determine the type of pavement maintenance and rehabilitation in a more consistent way.

Keywords: airport pavement management, crack density, pavement evaluation, pavement management

Procedia PDF Downloads 179

Authors: Yasuyuki Takahashi, Hayato Ishimura, Masao Miyagawa, Teruhito Mochizuki

Abstract:

Quantitative measurement of myocardium perfusion is possible with single photon emission computed tomography (SPECT) using a semiconductor detector. However, accumulation of ^99mTc-tetrofosmin in the liver may make it difficult to assess that accurately in the inferior myocardium. Our idea is to reduce the high accumulation in the liver by using dynamic SPECT imaging and a technique called time subtraction. We evaluated the performance of a new SPECT system with a cadmium-zinc-telluride solid-state semi- conductor detector (Discovery NM 530c; GE Healthcare). Our system acquired list-mode raw data over 10 minutes for a typical patient. From the data, ten SPECT images were reconstructed, one for every minute of acquired data. Reconstruction with the semiconductor detector was based on an implementation of a 3-D iterative Bayesian reconstruction algorithm. We studied 20 patients with coronary artery disease (mean age 75.4 ± 12.1 years; range 42-86; 16 males and 4 females). In each subject, 259 MBq of ^99mTc-tetrofosmin was injected intravenously. We performed both a phantom and a clinical study using dynamic SPECT. An approximation to a liver-only image is obtained by reconstructing an image from the early projections during which time the liver accumulation dominates (0.5~2.5 minutes SPECT image-5~10 minutes SPECT image). The extracted liver-only image is then subtracted from a later SPECT image that shows both the liver and the myocardial uptake (5~10 minutes SPECT image-liver-only image). The time subtraction of liver was possible in both a phantom and the clinical study. The visualization of the inferior myocardium was improved. In past reports, higher accumulation in the myocardium due to the overlap of the liver is un-diagnosable. Using our time subtraction method, the image quality of the ^99mTc-tetorofosmin myocardial SPECT image is considerably improved.

Keywords: 99mTc-tetrofosmin, dynamic SPECT, time subtraction, semiconductor detector

Procedia PDF Downloads 319

Authors: Chih-Wei Chao, Jiashing Yu

Abstract:

Microfluidic devices have recently emerged as promising tools for the fabrication of scaffolds for cell culture. To mimic the natural circumstances of organism for cells to grow, here we present three-dimensional (3D) scaffolds fabricated by microfluidics for cells cultivation. This work aims at investigating the behavior in terms of the viability and the proliferation capability of rat H9c2 cardiomyocytes in the gelatin 3D scaffolds by fluorescent images.

Keywords: microfluidic device, H9c2, tissue engineering, 3D scaffolds

Procedia PDF Downloads 411

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 116

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 149

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 106

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 178