Mobile Camera Research Articles

Development of a sustainable procedure for smartphone-based colorimetric determination of benzalkonium chloride in pharmaceutical preparations

A sustainable procedure offering green, simple, and rapid analysis was developed to determine benzalkonium chloride (BKC) in pharmaceutical preparations. The determination using smartphones was based on the ion pair colorimetric reaction with bromothymol blue (BTB), which produces a yellow color. The intensity of the product color, which is proportional to the concentration of BKC, was detected and evaluated using a smartphone camera and an image processing application. The procedure was performed in a microliter and was rapidly detected within 1 min after incubation. This offered high throughput at 28 samples per well plate in duplicate. Linear calibration, which was a plot of BKC concentrations and relative red intensities, was in the range of 2.0–24.0 μg/mL with an R2 of 0.997. The limits of detection (LOD) and quantitation (LOQ) were 1.0 and 3.2 μg/mL, respectively. This work was successful in applying it to pharmaceutical materials, disinfectant products, and pharmaceutical products containing BKC. It was discovered that the concentrations of BKC as an active ingredient in pharmaceutical materials were 82% w/v, whereas those in disinfectant products ranged from 0.4 to 2.1% w/v. In pharmaceutical products, ophthalmic drops and nasal sprays contain BKC as preservatives in the 0.01–0.02, and the 0.02% w/v, respectively. The results obtained by the proposed procedure compared with a reference titration method showed no significant differences at a 95% confidence level with 1.2–3.4% RSDs. This promotes the efficiency of pharmaceutical preparations regarding infection prevention and control by ensuring that available disinfectants contain a sufficient concentration of BKC. Additionally, this improves the efficiency of pharmaceutical preparations for quality control of pharmaceutical products by ensuring that the available preservatives maintain a sufficient concentration throughout the lifespan of the products.

Crop Leaf Disease Prediction Using Machine Learning

In India, crop area is largest in the world and produces major crops like wheat, pulses, fruits, rice and vegetables Despite of using modem taming techniques along with traditional, infectious plant diseases is major problem which can be caused by different viruses, fungus and bacteria. This mainly affects crop production as well as crop quality. It is very important to identify diseases at early stage Nowadays, automatic crop de detection has become a important research domain. It helps in detecting the symptoms of the disease when they are found on the e In this paper we will focus on finding the diseases in order to increase crop quality and production effectively. Here, we will focus on r diseases by observing leaves of plants at initial stage using machine learning. · In this paper, we designed a Deep Convolutional Neural Network based on LeNet to perform soybean leaf spot disease recognition and classification using affected areas of disease spots. The affected areas of disease spots were segmented from the leaves images using the Unsupervised fuzzy clustering algorithm. The proposed Deep Convolutional Neural Network model achieved a testing accuracy of 89.84%, and poor per class recognition results in 1378 images misclassified, and 1271 images correct classified. TheVGG16 achieved the best performance reaching a 93.54% success rate, and better per class recognition results in 1245 images misclassified, and 1404 images correct classified · In order to address the challenges related to the classification and recognition of soybean disease and healthy leaf identification, it is essential to have access to high-quality images. A meticulously curated dataset named “SoyNet” has been created to provide a clean and comprehensive dataset for research purposes. The dataset comprises over 9000 highquality soybean images, encompassing healthy and diseased leaves. These images have been captured from various angles and directly sourced from soybean agriculture fields; The soybean leaves images are organized into two sub-folders: SoyNet Raw Data and SoyNet Pre-processing Data.The SoyNet Pre-processing Data folder comprises resized images of 256∗256 pixels and the grayscale versions of disease and healthy images, following a similar organizational structure. We captured the images using the Nikon digital camera and the Motorola mobile phone camera, utilizing different angles, lighting conditions, and backgrounds. They were taken in different lighting conditions and backgrounds at soybean cultivation fields to represent the real-world scenario accurately. The proposed dataset is valuable for testing, training, and validating soybean leaf disease classification

Classification and Identification of Perfume Scents by an Enantioselective Colorimetric Sensor Array of Chiral Metal-Organic-Framework-Based Fabry-Pérot Films.

Many odors, like perfumes, are complex mixtures of chiral and achiral molecules where the cost-efficient (enantio-)selective sensing represents a major technical challenge. Here, we present a colorimetric sensor array of surface-mounted metal-organic-framework (SURMOF) films in Fabry-Pérot (FP) cavities. The optical properties of the FP-SURMOF films with different chiral and achiral structures are affected by the (enantio-)selective adsorption of the analytes in the SURMOF pores, resulting in different responses to the analyte molecules. The read-out of the sensor array is performed by the digital camera of a common smartphone, where the RGB values are determined. By analyzing the sensor array data with simple machine learning algorithms, the analytes are discriminated. After demonstrating the enantioselective response for a pair of pure chiral odor molecules, we apply the sensor array to detect and discriminate a large number (16) of common commercial perfumes and eau de toilettes. While our untrained human nose is not able to discriminate all perfumes, the presented colorimetric sensor array can classify all perfumes with great classification accuracy. Moreover, the sensor array was used to identify unlabeled samples correctly. We foresee such an FP-chiral-SURMOF-based sensor array as a powerful approach toward inexpensive selective odors sensing applications.

Development of a random forest based algorithm for road health monitoring

Road damages, such as potholes and rutting, must be addressed in their early stages to prevent accidents and minimize maintenance costs. The presence of these damages not only causes discomfort to passengers but also accelerates vehicle deterioration. Improper material mixtures and inadequate maintenance practices often contribute to road damage. In recent years, numerous machine learning (ML) algorithms have been developed with a predominant focus on binary classification, specifically targeting potholes. However, roads are susceptible to various damages beyond potholes, necessitating a comprehensive approach for effective solutions. This study introduces a novel algorithm utilizing a random forest (RF) classifier for multi-class classification of road damages. The proposed algorithm is validated through rigorous simulation and field studies. Unlike previous models, this approach embraces a comprehensive perspective by considering a wider range of damages, thereby facilitating a more nuanced and inclusive process for the classification of road damages. In the field study, data has been collected using a set of uni-axial accelerometers and a smartphone camera mounted on a test vehicle. The data has been processed using the sliding window technique, and features have been extracted from each window to train the RF classifier. In this process, an optimal window size has been determined and employed to enhance the effectiveness of feature extraction for training the RF classifier. The proposed algorithm demonstrates significant accuracy in both simulation and field studies, with notable performance in identifying and classifying road damages. The algorithm’s outcomes are utilized to estimate a detailed cost for repairing the identified damages. This paper showcases proficiency in addressing various damages, offering valuable insights for the implementation of cost-effective road maintenance strategies.

Soil Color Comparison Using Munsell Soil Color Chart and Calibrated Smartphone Camera

Soil color is a crucial property in soil fertility assessment and monitoring. However, the subjective nature of the Munsell Soil Color Chart (MSCC) can lead to uncertainty in the analysis. To address this issue, a study was conducted to develop a soil color classification model from smartphone digital imagery based on color analysis and MSCC. The study involved taking 26 soil samples from various soil types and locations in the Special Region of Yogyakarta, Indonesia. Digital images of the soil were taken through a smartphone camera and compared with observations using MSCC to compare color differences (ΔE) based on Lab values. Soil images obtained from indoor studio conditions and calibration using spydercheckr in indoor and outdoor conditions are compared with MSCC and Chromameter values. The L*a*b color space was found to be superior to RGB for predicting and detecting small differences in color. The study also found that the Munsell soil color chart (MSCC) had a lower color difference than the chromameter in all lighting conditions, indicating that the MSCC or visual assessment can better detect the main soil color or soil matrix, while chromameter readings may have errors due to soil impurities.

Smartphone-based device for rapid and single-step detection of piRNA–651 using anti-DNA:RNA hybrid antibody and enzymatic signal amplification

P-element-induced wimpy testis (PIWI)-interacting RNAs (piRNAs/piRs) are a class of small noncoding RNAs that play a crucial role in regulating various biological processes, including carcinogenesis. One specific piRNA, piR–651, has been reported to be overexpressed in both human blood serum and solid cancer tissues, that can be used a viable biomarker in cancer diagnosis. Early diagnosis of cancer can help reduce the burden of the disease and improve survival rates. In the present work, we report for the first time a smartphone-based colorimetric biosensor for highly sensitive and specific detection of piR–651 thanks to an enzymatic signal amplification, which yielded high colorimetric intensities. Indeed, a heteroduplex DNA:RNA was formed in the presence of piR–651 with the capture DNA probe immobilized on the magnetic beads for easy magnetic separation. Then, a HRP tethered to anti-DNA:RNA (S9.6) was used to reveal the DNA-RNA heteroduplex formed by catalyzing the oxidation of TMB substrate into colorimetric TMBox, which absorbs at 630 nm. The absorbance is positively proportional to the piR–651 concentrations. On the other hand, the colorimetric product of the assay can be photographed with a smartphone camera and analyzed using ImageJ software. Using a smartphone and under optimal conditions, the biosensor responded linearly to the logarithm of piRNA–651 from 8 fM to 100 pM with a detection limit of 2.3 fM and discriminates against other piRNAs. It was also successfully applied to the determination of piRNA–651 levels in spiked human serum.

Broadband achromatic and wide field of view metalens-doublet by inverse design

Metalenses, composed of patterned meta-atoms in various dimensions, offer tailored modulation of phase, amplitude, and polarization for diverse imaging applications across the visible and near-infrared spectra. However, simultaneously achieving achromatic and wide field of view (WFOV) imaging remains a significant challenge. In this paper, we propose a general inverse design framework for metalens-doublets that simultaneously enables broadband achromatic and WFOV imaging. The broadband achromatic and WFOV (BA&WFOV) metalens-doublet comprises a propagation phase metalens and a geometric phase metalens positioned on opposite sides of the substrate. This framework requires only once polarization conversion and mitigates aperture size constraints imposed by the limited group delay (GD) range of meta-atoms. We present a BA&WFOV metalens-doublet with an f-number of 3.9, a full field of view (FOV) of 68°, and a wavelength range from 640nm to 820nm. This metalens-doublet exhibits diffraction-limited focusing with an average absolute focusing efficiency of 16% and an average relative focusing efficiency of 60%. This innovative framework holds significant promise for applications in fields such as phone cameras, VR/AR, and endoscopes.

Create virtual dentoskeletal model by superimposing digital dental cast into cone-beam computed tomography scan.

Many artifacts and obstacles associated with cone-beam computed tomography (CBCT) scan can obscure or distort the details of the teeth and occlusal surface, like distorted teeth, streak artifacts, noise, and some malocclusion cases with excessive overlapping between jaws cause decrease the interocclusal space, which can impact diagnosis and treatment planning, and the 3D reconstruction accuracy. Optimizing dental precision by Integrating CBCT scans with other imaging modalities, supply more information to enhance CBCT accuracy, mainly in dental areas with limited clarity. Performing the Structure-from-Motion (SfM) photogrammetry method, using phone camera and photograph studio setup using simple hardware, to digitize the dental casts and obtain an accurate digital dental model. Using this digital dental model to enhance dental precision in the CBCT data by performing the superimposition process, using a surface-based registration method and integration process to create a virtual dentoskeletal model. Evaluate the accuracy and quality of the superimposition results using qualitative (visual inspection) and quantitative measures. The differences between the virtual dentoskeletal model and the reference CBCT model are calculating by the 3D Euclidean distance, the mean ± SD are 0.212 ± 0.169mm and 0.26 ± 0.149mm for the maxilla and mandible, respectively. The color-coded map shows that the two surfaces are similar, but the extremist values are concentrated in the dental region due to the presence of the noise in the reference model and the gingiva in the virtual dentoskeletal model. The resulting virtual dentoskeletal model can be viewed and manipulated on a computer screen, allowing for a detailed analysis of the teeth and supporting structures. The 3D model generated by the SfM photogrammetry technique did well during the superimposition process, representing a reliable method for virtual-based processing such as orthognathic surgery planning and splint design.

Evaluation of Taekwondo Poomsae movements using skeleton points

Taekwondo is a widely practised martial art and an Olympic sport. In Taekwondo, Poomsae movements are essential, as they form the foundation of the sport and are fundamental for success in competitions. The evaluation of Poomsae movements in Taekwondo has been a subjective process, relying heavily on human judgments. This study addresses the above issue by developing a systematic approach to evaluate Poomsae movements using computer vision. A long short-term memory-based (LSTM-based) machine learning (ML) model was developed and evaluated for its effectiveness in Poomsae movement evaluation. The study also aimed to develop this model as an assistant for self-evaluation, that enables Taekwondo players to enhance their skills at their own pace. For this study, a dataset was created specially by recording Poomsae movements of Taekwondo players from the University of Colombo. The technical infrastructure used to capture skeleton point data was cost-effective and easily replicable in other settings. Small video clips containing Taekwondo movements were recorded using a mobile phone camera and the skeleton point data was extracted using the MediaPipe Python library. The model was able to achieve 61% of accuracy when compared with the domain experts’ results. Overall, the study successfully achieved its objectives of defining a self-paced approach to evaluate Poomsae while overcoming human subjectivity otherwise unavoidable in manual evaluation processes. The feedback of domain experts was also considered to finetune the model for better performance.

نظام التشخيص المبكر للكشف عن نقص الفوسفور في حقول الشعير بناءً على شبكة YOLO-v8 المحسنة

This study presents a diagnostic method for early and rapid detection of phosphorus (P) deficiency in barley fields in the natural environment, stemming from computer vision and based on a deep learning approach. This is done by creating a training model based on the YOLOv8 network. A phenotyping method is used to identify the plant, and then based on the color of the leaves and stems to detect phosphorus (P) deficiency. Some improvements have been made to the basic algorithm, using image processing techniques such as flipping, rotating, deep convolution, and resizing for feature enrichment. In our approach, images were classified using polygonal bounding boxes in collaboration with agricultural experts to identify areas of interest for the model. The dataset was divided into validation, training, and testing. In the experimental phase, we analyzed the model's performance using a set of video clips captured with a mobile phone camera as a first stage, and our model achieved a detection accuracy of up to 80%. In the second stage, we tested the model using images and video clips taken from a group of cameras installed on top of the pivot irrigation machine to take advantage of the machine’s movement to scan the entire field. The model achieved a detection accuracy of up to 65%. Therefore, the proposed method can provide an early prediction system for plant needs, helping farmers maintain crop health and choose appropriate fertilizers. Keywords: Phosphorus (P), deep learning, YOLOv8, computer vision, image processing.

A comprehensive evaluation of marker-based, markerless methods for loose garment scenarios in varying camera configurations

In support of smart wearable researchers striving to select optimal ground truth methods for motion capture across a spectrum of loose garment types, we present an extended benchmark named DrapeMoCapBench (DMCB+). This augmented benchmark incorporates a more intricate limb-wise Motion Capture (MoCap) accuracy analysis, and enhanced drape calculation, and introduces a novel benchmarking tool that encompasses multicamera deep learning MoCap methods. DMCB+ is specifically designed to evaluate the performance of both optical marker-based and markerless MoCap techniques, taking into account the challenges posed by various loose garment types. While high-cost marker-based systems are acknowledged for their precision, they often require skin-tight markers on bony areas, which can be impractical with loose garments. On the other hand, markerless MoCap methods driven by computer vision models have evolved to be more cost-effective, utilizing smartphone cameras and exhibiting promising results. Utilizing real-world MoCap datasets, DMCB+ conducts 3D physics simulations with a comprehensive set of variables, including six drape levels, three motion intensities, and six body-gender combinations. The extended benchmark provides a nuanced analysis of advanced marker-based and markerless MoCap techniques, highlighting their strengths and weaknesses across distinct scenarios. In particular, DMCB+ reveals that when evaluating casual loose garments, both marker-based and markerless methods exhibit notable performance degradation (>10 cm). However, in scenarios involving everyday activities with basic and swift motions, markerless MoCap outperforms marker-based alternatives. This positions markerless MoCap as an advantageous and economical choice for wearable studies. The inclusion of a multicamera deep learning MoCap method in the benchmarking tool further expands the scope, allowing researchers to assess the capabilities of cutting-edge technologies in diverse motion capture scenarios.

Remote practical instruction using web browsers

<p>This paper introduces a novel approach to remote coaching, specifically targeting the body movements of learners participating remotely. The proposed system employs a smartphone camera to capture the learner’s body and represent it as a 3D avatar. The instructor can then offer guidance and instruction by manipulating the 3D avatar’s shape, which is displayed on a web browser. The main challenge faced by the system is to enable the sharing and editing of 3D objects among users. Since the HTML5 drag-and-drop feature is inadequate for transforming virtual objects consisting of multiple interconnected rigid bodies, the system tracks the pivot point of the manipulated rigid body. It assigns attributes such as pivot points and action points to each object, extending beyond their 2D screen coordinates. To implement the system, an interactive web application framework following the model-view-view-model (MVVM) architecture is utilized, incorporating Vue.js, Three.js, and Google Firebase. The prototype system takes advantage of the data binding capability of the framework and successfully operates within the 3D space of a web browser. Experimental results demonstrate that it can effectively share transformation information with an average delay of 300 ms.</p>

Xception transfer learning with early stopping for facial age estimation

The rapid development of deep learning attracts more attention to the analysis of person's face images. Deep learning methodsof facial age estimation are more effective compared to methods based on anthropometric models, models of active appearance, texture models, subspace of aging patterns. However, deep learning networks require more computing power to process images. Pre-trained models do not need a large training set and their training time is less. However, the parameters obtained as a result of transfer learning of the pre-training network significantly affect its efficiency. It is also necessary to take into account the properties of the processed images, in particular, the conditions under which they were obtained.Recently, the facial age estimation is implemented in applications in devices with limited resources of computing, for example, in smartphones. The memory size and power consumption of such applications are limited by the computing power of mobile devices. In addition, when photographing a person's face with a smartphone camera, it is very difficult to ensure the uniform lighting. The aim of the research is reducing the error of facial age estimation from uneven illuminated images by applying an early stopping of transfer learning of the Xception network. The proposed technique of transfer learning includes an early stopping of training, if the improvement of the results is not observed within a certain number of epochs. Then the network weights from the epoch with the lowest validation loss are saved. As a result of the proposed technique applying, the average absolute error of age estimation was about five years from unevenly illuminated test images. A number of parameters of the used in this case Xceptionnetwork is less than that of other deep learning neural networks which solved the age estimation problem. Then applying of the Xception network reduces the resource consumption of devices with limited computing power. Prospects for further research are reducing the unevenness of facial image lighting to decrease the error of age estimation. Also, to reduce the computing resources, it is promising to use fast transforms in the Xception convolutional layers.

Characterizing Low-cost Registration for Photographic Images to Computed Tomography.

Mapping information from photographic images to volumetric medical imaging scans is essential for linking spaces with physical environments, such as in image-guided surgery. Current methods of accurate photographic image to computed tomography (CT) image mapping can be computationally intensive and/or require specialized hardware. For general purpose 3-D mapping of bulk specimens in histological processing, a cost-effective solution is necessary. Here, we compare the integration of a commercial 3-D camera and cell phone imaging with a surface registration pipeline. Using surgical implants and chuck-eye steak as phantom tests, we obtain 3-D CT reconstruction and sets of photographic images from two sources: Canfield Imaging's H1 camera and an iPhone 14 Pro. We perform surface reconstruction from the photographic images using commercial tools and open-source code for Neural Radiance Fields (NeRF) respectively. We complete surface registration of the reconstructed surfaces with the iterative closest point (ICP) method. Manually placed landmarks were identified at three locations on each of the surfaces. Registration of the Canfield surfaces for three objects yields landmark distance errors of 1.747, 3.932, and 1.692 mm, while registration of the respective iPhone camera surfaces yields errors of 1.222, 2.061, and 5.155 mm. Photographic imaging of an organ sample prior to tissue sectioning provides a low-cost alternative to establish correspondence between histological samples and 3-D anatomical samples.

Traffic Counting using YOLO Version-5 (A case study of Jakarta-Cikampek Toll Road)

The Jakarta-Cikampek toll road is the main access to the Tanjung Priok port, which is connected directly via the Cilincing-Tanjung Priuk Port toll road as a development of the North Jakarta reclamation coastal area. YOLO (You Only Look Once) is a common object detection model that offers faster and more accurate results.. The purpose of this article is to use advancements in information technology to automate the process of manually recording traffic counts on the highway. The method utilized in this study was to record a video of traffic movements with a smartphone camera and save it in MP4 format. Calculations are performed at the office after receiving recorded video and utilizing a program written by the author that makes use of Python, OpenCV, Pytorch, and YOLO version 5 software. When passing through a counter box, the traffic volume is counted and saved in Excel format (.xls). The video records footage near the Tambun area of the Jakarta-Cikampek toll road. According to the measurement accuracy of 95% for cars, 96% for buses, and 89% for trucks respectively, it can be stated that using YOLO version 5 for detecting vehicle volume and categorization is fairly satisfactory.

Method for Human Ear Localization in Controlled and Uncontrolled Environments

One of the fundamental stages in recognizing people by their ears, which most works omit, is locating the area of interest. The sets of images used for experiments generally contain only the ear, which is not appropriate for application in a real environment, where the visual field may contain part of or the entire face, a human body, or objects other than the ear. Therefore, determining the exact area where the ear is located is complicated, mainly in uncontrolled environments. This paper proposes a method for ear localization in controlled and uncontrolled environments using MediaPipe, a tool for face localization, and YOLOv5s architecture for detecting the ear. The proposed method first determines whether there are cues that indicate that a face exists in an image, and then, using the MediaPipe facial mesh, the points where an ear potentially exists are obtained. The extracted points are employed to determine the ear length based on the proportions of the human body proposed by Leonardo Da Vinci. Once the dimensions of the ear are obtained, the delimitation of the area of interest is carried out. If the required elements are not found, the model uses the YOLOv5s architecture module, trained to recognize ears in controlled environments. We employed four datasets for testing (i) In-the-wild Ear Database, (ii) IIT Delhi Ear Database, (iii) AMI Ear Database, and (iv) EarVN1.0. Also, we used images from the Internet and some acquired using a Redmi Note 11 cell phone camera. An accuracy of 97% with an error of 3% was obtained with the proposed method, which is a competitive measure considering that tests were conducted in controlled and uncontrolled environments, unlike state-of-the-art methods.

Properties of processing and data acquisition methods in test methods of chemical analysis

Chemical analysis is a crucial source of information for assessing the state of the environment, evaluating the human body condition, and serving as a diagnostic tool for detecting diseases and more. The use of simple, rapid, and cost-effective analysis methods without the need for specialized laboratory conditions and personnel with specific qualifications, such as test methods, contributes to the effective implementation of well-known techniques for analyzing the chemical composition of environmental objects and for use in the field of healthcare. Scanner indication of the analytical signal in chemical analysis is an objective, fast, and automated method for assessing the color characteristics of colored samples. The use of scanner indication in obtaining the analytical signal eliminates errors associated with subjective evaluation and assists individuals with color vision impairments, speeding up data collection and processing operations. Considering and controlling the light source in scanning systems is crucial to ensuring the stability and accuracy of obtained images under different lighting conditions. To obtain the analytical signal, it is primarily necessary to process the image to obtain the region of interest, which is determined by segmentation, manual methods, and using image processing techniques. The analysis of obtained values is carried out by constructing a linear or nonlinear relationship of color values in the selected color model converted to a single value of the analytical signal to concentration. Most existing systems for scanner detection of the analytical signal in analytical test methods require human involvement in obtaining and interpreting results, limiting the application areas of these methods of analytical signal indication. Further research is focused on creating a reliable method for obtaining the analytical signal using a smartphone camera based on image processing techniques, allowing for the automation of obtaining, processing, and improving the reproducibility of results in chemical test analyses.

Augmented Reality-Based Product Video Promotion Application: A Framework and Implementation

Abstract: Augmented Reality (AR) technology has gained significant attention in recent years due to its potential to enhance user experiences across various domains. In the realm of marketing and promotion, AR presents a promising avenue for engaging consumers in immersive product experiences. This paper proposes a framework and implementation of a product video promotion application leveraging augmented reality technology. The application aims to revolutionize traditional product promotion strategies by providing users with interactive and immersive experiences through AR-enhanced product videos. The proposed system integrates computer vision, 3D modelling, and AR rendering techniques to overlay virtual product information seamlessly onto the real-world environment captured through a smartphone or tablet camera. Additionally, the application incorporates features such as real-time product visualization, interactive product demonstrations, and social sharing functionalities to enhance user engagement and promote viral marketing. Through a comprehensive review of related works and the presentation of our framework and implementation details, this paper contributes to the growing body of research on ARbased marketing solutions and provides insights into the potential of AR technology for transforming product promotion strategies.

Artificial intelligence enabled smart phone app for real-time caries detection on bitewing radiographs.

Diagnosis of proximal caries is a difficult task. Artificial intelligence (AI) enabled diagnosis is gaining momentum. Therefore, it is of interest to evaluate the effectiveness of an artificial intelligence (AI) smart phone application for bitewing radiography towards real-time caries lesion detection. The Efficient Det-Lite1 artificial neural network was used after training 100 radiographic images obtained from the department of Oral Medicine. Trained model was then installed in a Google Pixel 6 (GP6) smartphone as artificial intelligence app. The back-facing mobile phone video camera of GP6 was utilised to detect caries lesions on 100 bitewing radiographs (BWR) with 80 carious lesion in real-time. Two different techniques such as scanning the static BWR on laptop with a moving mobile and scanning the moving radiograph on the laptop with stationery mobile were used. The average value of sensitivity/precision/F1 scores for both the techniques was 0.75/0.846 and 0.795 respectively. AI programme using the rear-facing mobile phone video camera was found to detect 75% of caries lesions in real time on 100 BWR with a precision of 84.6%. Thus, the use of AI with smart phone app is useful for caries diagnosis which is readily accessible, easy to use and fast.

Differentiate the Varieties of Plants Using Leaf Images with YOLOv4

There are many types of plant, not all recognized by people. Indonesia has many varieties of plants across the country, from Sabang to Merauke. Often, we find some fruits or plants that grow in different areas called by different names. The goal of this research is to create an application that can recognize plant types from their characteristics. These characteristics can be leaves, fruit, flowers or roots. Current research focuses on distinguishing several types of plants including their varieties using leaf images. It is hoped that apart from being able to determine the mango plant, we can also determine the type of mango, such as manalagi, harum manis, as well as guava and orange. To be able to recognize these types of plants, an object recognition algorithm based on CNN (Convolutional Neural Network) will be used. The algorithm that will be used is YOLO (You Only Look Once) version 4. Method in this research includes data collection, all data was taken using webcam or mobile phone camera. The next step is data preparation and creating a bounding box. Next step is creating model using training data and the last step was compare model with testing data, The result we got is really encouraging with almost 100 percent accuracy. The lowest accuracy is 98 percent for jambu air. This research show us that YOLOv4 can be used to differentiate varieties of plants.