Vitamin Deficiency Identification using Image

Processing
1st Nivedita Shimbre 2nd Prema Sahane 3rd Sarim Ansari
Computer Department Computer Department Computer Department
JSPM’s Rajashi Shahu College of JSPM’s Rajashi Shahu College of JSPM’s Rajashi Shahu College of
Engineering,Pune-411033 Engineering,Pune-411033 Engineering,Pune-411033
Maharashtra,India Maharashtra,India Maharashtra,India
nnshimbre_comp@jspmrscoe.edu.in pbsahane_comp@jspmrscoe.edu.in sarimmohd652@gmail.com

4th Rutuja Gadhave 5th Sakshi Jagdhane 6th Ashutosh Sharma

Computer Department Computer Department Computer Department
JSPM’s Rajashi Shahu College of JSPM’s Rajashi Shahu College of JSPM’s Rajashi Shahu College of
Engineering,Pune-411033 Engineering,Pune-411033 Engineering,Pune-411033
Maharashtra,India Maharashtra,India Maharashtra,India
rutujagadhave19@gmail.com sakshijagdhane11@gmail.com ashutoshsharmam200@gmail.com

Abstract—Vitamin deficiency significantly impacts overall are the effective treatment. Traditional methods of diagnosis
health, leading to anemia, complications during pregnancy or are checking the blood and the urine in the labs. Different tests
Vitamin deficiencies are a significant public health concern in the labs are very time consuming are expensive instead of
worldwide, particularly in resource-constrained settings. Early doing this we can identify the vitamin deficiency simply from
detection and diagnosis of vitamin deficiencies are crucial for
timely intervention and prevention of long-term health
the captured images of our organs. This approach of
consequences. Traditional methods of diagnosis is time- identification is cost effective as well as time efficient.
consuming and more costly which includes blood tests. Recent
advancements in image processing and computer vision gives
Vitamins are important micronutrients that the body requires
cost-effective screening of vitamin deficiencies childbirth, and for various physiological processes. Early detection of vitamin
impaired cognitive and physical development. It is essential to deficiencies makes the user’s work easier and saves time and
detect and treat such deficiencies to avoid adverse effects. money. Blood tests are standard practice in rich countries, but
Traditionally, vitamin deficiency is identified through blood different socio-economic and public perspectives make testing
tests, which is costly and takes time also. Consequently, many difficult in poor countries[5].
individuals neglect their vitamin health. Our objective is to
provide a time-efficient and cost-free method to check for
vitamin deficiencies at any time. Vitamin deficiencies show This survey study offers through summary of previous studies,
visible impacts on organs such as the eyes, nails, lips, tongue, provides a comprehensive overview of existing research on
and skin. We aim to identify these impacts and determine vitamin deficiency identification using image processing
specific vitamin deficiencies through images of various organs techniques, including image acquisition, processing, feature
using advanced image processing techniques. We use extraction, and the classification methods. The use of image
sophisticated deep learning techniques and machine learning processing techniques and CNNs allows for automatic
algorithms to identify vitamin deficiency efficiently. After analysis of visual cues in various body parts, enabling the
identification of deficiency, our system will suggest dietary
detection of multiple types of vitamin deficiencies[15]. This
recommendations to address the issue. If the deficiency is severe,
it will also advise consulting a doctor. Hence, we propose a software is a tool that helps address the global problem of low
system to facilitate early diagnosis of vitamin deficiencies, saving nutritional awareness and helps medical professionals make
time and costs and making it convenient for users. more accurate diagnoses.

Keywords—Vitamin Deficiency, Image Processing,

Convolutional Neural Network(CNN), early detection, machine II. RELATED WORK
learning, accuracy, Artificial Neural Network(ANN) . In this paper, we are surveying previous work related to
vitamin deficiency identification. This paper uses various
I. INTRODUCTION machine learning algorithms for their implementation such as
Approximately, 2.0 billion people are deficient in different CNN, ANN, KNN, etc.
vitamins. Nearly 1.20 billion people who are deficient in zinc,
half million people die each year. In the United Arab Emirates, To identify vitamin deficiencies, this model uses deep
almost 91% of the population suffers from deficiencies in learning methods including NASNetMobile, Convolutional
various multiple vitamins. Neural Networks (CNN), Inception, Artificial Neural
Networks (ANN), and MobileNet. The dataset used is of
Every year many people suffer due to vitamin deficiency and images of bluish nails, lips, tongues, and eyes damaged by
may get attacked of severe disease. One-third of the global glaucoma that were taken from publicly accessible sources
population, is reported as the Vitamin Deficient according to and Google Photos. Following an analysis of these methods
the latest research of the World Health Organization.Vitamin performances, MobileNet was shown to be the most effective,
deficiency affects on the health of the human being causes the obtaining 96% training accuracy and 97.2% validation
symptoms like mild fatigue. Early detection and the diagnosis accuracy[1].
The model uses a set of images of the lips ,tongue, nails, and calcium, potassium, vitamin D, iron, and Vitamin B1 (with or
eyes to identify vitamin deficiencies. Given the widespread without anemia) [8].
use of CNNs in fields like image classification and medical
image analysis, a CNN model was constructed and trained The method for identifying vitamin deficiencies presented in
using TensorFlow and Keras. OpenCV was used to handle this paper makes use of CNN and image processing
and process fresh images for testing and validation, techniques. The method comprises preprocessing, feature
guaranteeing the efficiency and accuracy of the model [2]. extraction using a CNN that has already been trained, image
acquisition, and classification using patterns that have been
The main aim of this project is to apply image analysis and learned. Images of the tongue, skin, hair, and nails are used
deep learning techniques to detect vitamin deficiencies. They as parameters to identify vitamin deficiencies. The suggested
gathered a huge amount of dataset and preprocessed them to method, which makes use of deep learning and computer
enhance features. These images will be preprocessed to vision, provides a simple, accurate, and non-invasive means
enhance features and decrease noise. They used AlexNet of identifying and tracking dietary deficits [9].
,DNN algorithm for their implementation [3].
They created an AI-driven system for identifying vitamin
Using input data such height, weight, age, sex, BMI, milk deficiencies as part of this study. They used pictures of lips,
consumption, waist circumference, bone mass, and sunlight tongue, nails, and eyeballs for this. In order to extract
exposure, vitamin D insufficiency was found in this study. particular features and qualities from the photos, they
This was accomplished by combining the LightGBM and Cat integrated machine learning approaches, and then they used a
Boost algorithms, which greatly improved the diagnostic Fuzzy Logic to determine the type of the deficiency.[10].
process. The integration of SHAP (Shapley Additive
Explanations) enhanced the usefulness of the model. The 481 observations from people who voluntarily applied to the
model's precision was 96.3%, and its accuracy was roughly Internal Medicine Department were included in the study.
95.3% [4]. Four traditional supervised machine learning models were
tested for classification performance: Random Forest (RF),
Detection of vitamin deficiencies by examining images of Support Vector Machine (SVM), Elastic-Net Ordinal
nails, skin, and hairs, utilizing CNN algorithms for detection. Regression (ENOR), and Ordinal Logistic Regression (OLR).
The use of the AlexNet algorithm in image classification is This model performed comparative analysis on classification
emphasized, though challenges like variations in image of individual’s metabolic syndrome positive status,hyper-
quality and the need for strong algorithms are recognized. parametric tuning [11].
This approach aims to enhance accessibility and efficiency in
diagnosing vitamin deficiencies, offering potential Shows the overall study of Convolutional Neural Networks
advancements in telemedicine [5]. and transfer learning primarily for medical imaging. Transfer
learning involves Pretrained CNN models which will be
significant to deal with challenges like small dataset, limited
Uses pictures of the tongue, nails, and eyes to identify vitamin computational resources. They have used CNN structures
deficiencies. These photos are classified using Deep Learning like DenseNet,LeNet-5, VGG-1.6 Net, ResNet, AlexNet, etc.
based on color and texture, which yields likely diagnoses. This paper give important overview about which structure
The approach makes use of the Adam optimizer to get MSE should be use for particular type of project[12].
(Mean Squared Error) findings that are more accurate.
Utilizing the CNN (Convolutional Neural Network) deep It is related to identification and classification of plant leaf
learning technology, image parameters are better understood. diseases of beans plant. For this, it uses a publicly available
The accuracy rate is more than 90% [6]. dataset having 1296 images of bean leaves having images of
bean leaves including healthy leaf and unhealthy leaf images.
The goal of the study is to identify vitamin A insufficiency in For identification, a deep learning approach is implemented
students. A small dataset of visual symptoms and diagnoses using MobileNet model and Tensorflow Open source library.
get from routine eye exams in Maradi, Niger, in the school MobileNetV2 and Adam optimizer tool is used to train the
setting, was subjected to a variety of machine learning proposed model. The training, testing and validation dataset
algorithms. By utilizing only current medical records, the sizes were 80%, 10% and 10% with respect to the used
researchers hoped to develop a controlled to detect vitamin A dataset giving an accuracy of the proposed model as 97%,
deficiency in children of school that would be accessible, 92% and 94.74% on the training dataset, test dataset and
inexpensive, and successful. The model with the highest validation dataset respectively[13].
accuracy, the LGB model, has 84.4% accuracy rate, 81.9%
sensitivity, and 84.7% specificity [7]. The developed system is an android application. System takes
images of lips, tongue, nails and eyes. This system performs
By use of different machine learning algorithms and vitamin deficiency detection using Natural Language
retrospective files from a tertiary care hospital, the Processing (NLP), Fuzzy membership function &
researchers identified vitamin C insufficiency. They used Defuzzification, MATLAB and Central Neural Networks
Bayesian networks and logistic regression for analysis. They (CNN) and recommends food supplements accordingly. They
come to the conclusion that testing for vitamin C deficiency have made the android app which have less software support
is necessary if a patient has combined deficiencies of iron, and users may face various types of issues [14].
They aims to get the best machine learning algorithm or hydroxyvitamin D. Compared to typical ELISA test kits, the
prediction of vitamin B deficiency. Primarily, data was platform's colorimetric detection method, which uses an
collected from 3040 college students having age between 18- immunoassay based on gold nanoparticles, having precision
21. This dataset uses various input parameters like milk better than 10 nM and accuracy of 25 nM.[21].
consumption, body fat, sunlight exposure , waist
circumference, height, age, sex, BMI, bone mass, Exercise,
etc .For accuracy analysis algorithms like Random Forest, The paper focuses on creating an expert system that
SVM, LDA, Ada Boost, MLP, KNN classifier were used. diagnoses vitamin and mineral deficiencies based on user
Among these, the Random classifier having the highest symptoms by utilizing a neural network. The goal is to offer
accuracy of 96% [15]. a tool that makes it easier to identify and treat nutritional
imbalances early on, maybe leading to an improvement in
Technique identifies deficiency of fat-soluble Vitamins A, D, general health. The technology maps symptoms to
E and K. Primarily, data was collected from previous works inadequacies using rules and pre-existing datasets for
and nutritional experts. Technique Backpropagation Neural training. The model's fundamental operations involve
Network is used. The system takes “yes” or “no” type value learning from training examples, representing knowledge,
for a list of 30 symptoms and on this basis probable vitamin and implementing a neural network. The system accuracy is
deficiency is predicted. The system achieved an accuracy of 60%[22].
82 % [16].
Detection of Diabetic Retinopathy using Retinal images. This
The system is an Android mobile application and performs dataset includes retinal images of left and right eye having
high resolution and different imaging conditions. Diabetic
analysis of vitamin deficiency on the basis of provided
Retinopathy get evaluated according to the various scales.
images of tongue, nails, lips and eyes. The dataset for this
Analysis of images is done on the basis of 512 x 512 pixels
project was taken from internet resources. Techniques used
are GUI, CNN, Sigmoid Squishification Weight Function, resolution input. [23].
fuzzy membership function using MATLAB, Android studio,
Tensorflow and defuzzification. After uploading the images, Identifies the vitamin and mineral deficiencies including
to check accuracy of detection user have to answer three vitamins and 6 minerals. They collect information from
questions.[17]. experts and identify 46 symptoms. They set 35 rules based on
this information of symptoms using decision tables. This
A brand-new machine learning algorithm called "Seleukos" system uses forward-chaining methods to deal with the
is presented in the paper with the goal of enhancing prediction uncertainties that arise due to the Mycin certainty factor. This
accuracy and reducing the learning curve for very big study successfully made a system using the method of
datasets. The main goal is to improve upon the lr-partition calculation certainty factor (CF) to defeat the uncertainty
algorithm by resolving its shortcomings in handling factor which helps for vitamin and mineral deficiency
undefined combinations of attribute values, hence enhancing identification in the human body.[24].
the classification accuracy of current approaches.. This
system proposes to change the code to optimize and Proposed system is an expert system implemented using
investigate to increase program efficiency[18]. decision tree technique to diagnose the vitamin deficiencies
on the basis of the given answers to questions about
Study of different cnn algorithms for face skin disease symptoms that the user is facing and the conclusion is
classification using clinical images. They uses five CNN described using depth first search (DFS) engines and forward
mainstream structures like Resnet50, Inception V3, chaining method. This system can diagnose for vitamin
Densenet-121, Xception, Inception Resnet V2. Over 6 face deficiencies such as vitamin A, D, C, K and B1, B2, B3, B5,
skin diseases like BCC, LE, ROS, SK, AK, SCC were taken B6, B7, B9, B12 [25].
into account by them. Also they suggested one strategy to
work on skin diseases that first train the model using images
of whole body parts and then retrain model for specific body
part[19]. III.OBSERVATION TABLE
Table 1 summarizes the following papers on Vitamin
Project tries to implement a CNN model that gets supported Deficiency detection techniques using GUI, CNN, Random
on eco-friendly devices. It uses MobileNet v1 architecture Forest (RF), Support Vector Machine (SVM),etc. with
and depthwise separable convolutions technique of CNN provided title, year ,dataset, performance metric, and
using Tensorflow open source library. For training and challenges. Overall the review provides a good landscape of
testing purpose, caltech101 dataset is used and the resulted the research done.
accuracy is 92.4% [20].

The research details the creation of a portable platform that

uses a paper-based microfluidic device to quickly quantify
vitamin D levels. The goal is to develop an affordable, easily-
accessible diagnostic tool that uses a smartphones and a test
strip to precisely evaluate vitamin D levels, particularly 25-
 TABLE 1:COMPARISION TABLE
Sr. Title Year Methodology Challenges
No

1. Vitamin Deficiency 2024 ANN,CNN, They collected images for the dataset from publicly accessible
Detection using Neural InceptionV3, platforms and Google Photos instead of varied platforms like
Network NASNetMobile, hospitals. They did not build mobile application or website for
MobileNet. users to access their model to test for vitamin deficiency. The
MobileNet- effects of vitamin deficiencies on nails and eyes cannot be
Accuracy fully understood by relying just on blue nails and eyes affected
=97.2% by glaucoma.

2. Vitamin deficiency 2024 CNN As this is a medical undertaking, training the model will require
detection using image a large number of photos. However, it might be difficult and
processing and neural complex to train a CNN model on such big datasets.
network

3. Novel method 2024 AlexNet, They did not use a more sophisticated and accurate algorithm
vitamin deficiency for analyzing images.
DNN
detection using
ALEXNET DNN
algorithm.

4. Vitamin ‘D’ Deficiency 2024 LightGBM and A model's capacity to be generalized may be greatly impacted
Detection: A novel CatBoost by biases and constraints in the dataset. Prior to implementing
ensemble Accuracy 95.3% the model in actual medical settings, ethical issues and
approach with resource limitations must also be resolved, particularly when
interpretability insights. combining it with Internet of Things sensors to enable real-
time monitoring.

5. Vitamin Deficiency 2024 CNN, AlexNet One of the challenges is that hair is employed as a parameter
Detection for vitamin detection, thus a diverse dataset is required.
Using image processing
and Deep-CNN Additionally, since sensitive health information is being used,
Algorithm. it is imperative to protect privacy and have patient consent.

6. Vitamin Deficiency 2024 MSE,CNN Use of images of only the tongue, nails, and eyes, which do not
Detection cover all the potential impacts of vitamin deficiency.
Accuracy 90%
Using Convolution
Neural
Network with adam
optimization.

7. Detecting vitamin 2023 LGB model Limited demographic factors for identifying vitamin A
‘A’ Deficiency in school deficiency, despite the fact that vitamin A has a significant
Accuracy 84.4%
children Using impact on children's bodies
an enhanced explainable
ML Model

8. Using ML Models to 2023 Bayesian Limited Dataset

Predict networks and
Vitamin ‘C’ deficiency logistic
in patients with regression
metabolic and bariatric
surgery
Sr. Title Year Methodology Challenges
No

9. Vitamin deficiency 2023 CNN Use of CNN algorithm which can be complex for large
detection using Image dataset.
Processing and neural
network

10 Vitamin deficiency 2023 Fuzzy Logic There is limitation of vitamin deficient people hence model
detection Using image decision-making was not directly tested on vitamin deficient people.
Processing and Neural system
Network

11. ML approach for the 2023 Random Forest The restriction of small sample size.
detection (RF), Ordinal
of vitamin ‘D’ level: a Logistic
Comparative Study Regression
(OLR).

12. A Study of CNN and 2023 LeNet-5 Limited Quantity of dataset.

Transfer Learning in (accuracy-
Medical Imaging. 99.2),Resnet,
DenseNet
Alexnet.
13. Beans leaf diseases 2022 jan Tensorflow, The model is trained on only 3 categories which includes 2
classification using MobileNetV2, major types of diseases so it doesn’t covers other minor
MobileNet models CNN, GPU, diseases related to beans plant.
adagrad
optimizer, nadam
optimizer, SGD
optimizer,
RMSprop
optimizer, Adam
optimizer
14. Vitamin deficiency 2021 Natural Langauge The efficiency of the system can be increased by adding more
Detection using Image Processing(NLP), data with the help of doctors and experts.
Processing and AI Fuzzy
membership Android applications have limited software support and users
function, CNN may face compatibility issues.

15. A predictive 2020 Random Forest, Not validated for all age groups.
performance analysis of SVM, LDA
Vitamin ‘D’ Deficiency Accuracy 84.4% Random forests are prone to overfitting if the data contains a
severity Using ML large number of features and very slow in making predictions
Methods when compared to other algorithms.

16. Identification of fat- 2020 Backpropagation Limited dataset is used.

soluble vitamins Neural Network Unclean or noisy data can affect the backpropagation algorithm
deficiency using ANN. Accuracy 82% while training a neural network.

17. Vitamin deficiency 2020 GUI, CNN, Dataset used is limited and not well-authenticated.
detection using Image Sigmoid Detection accuracy can be affected if the given answers to
Processing and Neural Squishification asked questions are false or improper.
Network Weight Function
Minimum image resolution should be of 439 x 335 pixels.
Sr. Title Year Methods Used Challenges
No

18. A Novel ML Algorithm 2019 C4.5, SVM, and Subsequent research proposes changing some of the code for
to Reduce prediction KNN optimization and investigating additional improvements to
error and accelerate increase the efficiency of the program.
Learning curve for very
large datasets.

19. Studies on Different 2019 Resnet50-(Recall Application of AI in medical field is not sufficient
CNN Algorithms for 63%), Inception V3-
Face Skin Disease (Recall 66%), Dataset should be improved by quality and quantity wise.
Classification Densenet-121
Based on Clinical (Recall 68
Images. %),Xception-Resnet
V2-(Recall 77%)

20. Efficient mobilenet Oct 2019 CNN, MobileNet, The depth wise separable convolutions technique may
architecture as image depthwise separable accuracy poor direct information communications, resulting
recognition on mobile convolutions, in loss of accuracy.
and embedded devices GEMM
Optimizing techniques are not used to increase the accuracy
of the implemented model.

21. A portable platform for 2018 Arduino, Cloud, Increasing accuracy and precision, especially for applications
the quantification of TCS230 using genuine samples, like testing of human serum is
vitamin ‘D’ levels by required.
using paper-based
Microfluidic

22. Vitamin and minerals 2018 Artificial Neural Enhancing the precision of symptom mapping and
Diagnosis System Using Network, Expert improving the neural network's learning algorithms to
Neural Network system, increase diagnostic accuracy is required.
Accuracy 60% System accuracy is low.

23. Detecting diabetic 2017 CNN, CUDA Deep There is variations in discrete metrics and small classes which
retinopathy from retinal Neural Network leads to misclassification.
Images Using CUDA Accuracy 85%
DNN.

24. An expert system for 2016 Artificial Neural Knowing the vitamins deficiencies and their percentage in
early diagnose of Network, Certainty body is required by the patient.
vitamins and minerals Factory, Expert
deficiency System. Expert guidance is required to give input as patient may not
on the Body. be aware about all the symptoms of a specific vitamin
deficiency.

25. The expert ystem 2015 Depth First Search It is a desktop-based software hence population of
application for algorithm, Forward accessible users can be limited.
Diagnosing Human Chaining algorithm,
Vitamin Deficiency Artificial Neural The accuracy depends on the correctness of the user inputs.
Through forward Network, Decision
chaining methods Tree Diagram.
.
 IV.METHODOLOGY
Vitamin deficiency detection uses different methodolgy to
detect correct vitamin deficiency. Here are the some
methodologies and concepts associated with it.
1. Machine Learning:
Within the field of AI, machine learning builds models and
generate algorithms that helps computer to analyze and make
judgments. It entails teaching machines to see trends,
anticipate outcomes, and gradually enhance their capabilities
without being specifically designed for certain jobs.

i.Neural Network:
A neural network in machine learning is a statistical model
motivated by human brain, including layers of
interconnected nodes that process and transmit details.
These Networks learn to perform tasks like classification,
prediction, and pattern recognition by adjusting the Fig2: Comparision graph for accuracies of different
connections (weights) between nodes based on the data they Machine Learning Techniques.
receive, improving their accuracy over time.
Among the worked-out techniques, it has been shown that
2.Deep Learning: DL is a branch of ML that copy how the Mobilenet and KNN stand out as one of the most accurate
human brain processes the massive quantities of data techniques having approximately 97% and 96% accuracy
through several layers of interconnected nodes in order to respectively. Besides these, other techniques like ANN,
teach computers to see patterns and make judgments. For NASNetMobile, and CNN have accuracies above 90%.
tasks like picture identification, speech processing, and LightGBM has an accuracy of 89.2% whereas InceptionV3
natural language understanding, this method is particularly gave the lowest performance in terms of accuracy, 79.1%.
effective.
V.DATASET
2.1 Image Processing:
It is a technique that is use to enhance, analyze, and The proposed model works on the Vitamin Deficiency Dataset
to capture images. Open Computer Vision(Open CV ) library
manipulate images using computer algorithms. It involves
is used to produce our dataset.. There is limited work done on
operations like filtering, resizing, and transforming images to
the Vitamin Deficiency Dataset, therefore our aim is to
enhance their quality, gains information and use them for enhance the correctness of the proposed model by using the
further analysis in tasks such as object recognition or medical well-defined dataset. After capturing images processing is
imaging. done by using a ML algorithm to generate the output
3.Quantitative Analysis Organ Image Description
3.1. Performance Metrics Lips Detect types of Vitamin
Precision is the ratio of true positive values of vitamins to the B deficiency from
total positive values. Recall tells how many actual positive Cracked lips (cheilosis)
values are predicted Accuracy is the ratio of correctly Eyes We can get deficiency
predicted values of vitamins to the total values of vitamins. of vitamin A from eyes.
F1 score is a weighted average of recall and precision(vitamin).
Skin Dry and lackluster skin
Precision = TPVitamin detects the Vitamin D
(TPVitamin+ FPVitamin) deficiency. Rough dry,
and scaly skin, detects
Recall = TPVitamin the Vitamin C
(TPVitamin +FNVitamin) deficiency.
Tongue A magenta-colored
Accuracy = TPVitamin +FNVitamin tongue, soreness, and
(TPVitamin +FNVitamin+TNVitamin+FPVitamin) swelling denotes the
riboflavin Vitamin(B2)
F1-Score = 2*(Recall*Precision)Vitamin deficiency
(Recall + Precision)Vitamin Nails Brittle nails that are
where, prone to splitting and
TPVitamin= True Positive of Vitamin breaking and denotes
FPVitamin=False Positive of Vitamin the vitamin B7
TNVitamin= True negative of Vitamin deficiency.
FNVitamin= False Positive of Vitamin Fig 3: Dataset for identification of vitamin deficiency
 VI.CONCLUSION [9] Dr. R. Maruthamuthu, T. Harika, "Vitamin Deficiency Detection
Using Image Processing and Neural Network", International Journal
In summary, Vitamin Deficiency Detection system constitute of Scientific Research in Computer Science, Engineering and
to detect the vitamin deficiency in a simple and easy way. The Information Technology (IJSRCSEIT), ISSN : 2456-3307, Volume 9
Issue 4, pp. 200-205, July-August 2023 doi :
survey has provided a comprehensive overview of the https://doi.org/10.32628/CSEIT23903112
vitamin deficiency identification using image processing [10] Rutuja Moholkar, Mansi Kamble, Gauri Bobade, Saijyoti Shinde
techniques. The reviewed studies demonstrate the potential of “Vitamin Deficiency Detection Using Image Processing and Neural
image processing and computer vision in detecting vitamin Network”,2023..https://ijariie.com/AdminUploadPdf/Vitamin_Deficie
ncy_Detection_Using_Image_Processing_and_Neural_Network_ijarii
deficiencies and diet recommendation. Future research e20832.pdf
directions includes of Exploring deep learning and transfer [11] Sancar, Nuriye & Tabrizi, Sahar. (2023). Machine learning approach
learning techniques for improved accuracy, Investigating for the detection of vitamin D level: a comparative study. BMC
multimodal imaging approaches for comprehensive Medical Informatics and Decision Making. 23. 10.1186/s12911-023-
assessment. By addressing these challenges and leveraging 02323-z.
advancements in image processing and computer vision, we [12] Salehi, A.W.; Khan, S.; Gupta, G.; Alabduallah, B.I.; Almjally, A.;
Alsolai, H.; Siddiqui, T.; Mellit, A. A Study of CNN and Transfer
can move towards non-invasive, accurate, and accessible Learning in Medical Imaging: Advantages, Challenges, Future
screening tools for vitamin deficiencies, ultimately to Scope. Sustainability 2023, 15, 5930.
improve public health and quality of life. https://doi.org/10.3390/su15075930
In the future, our focus will be on the accuracy of the dataset [13] E. Elfatimi, R. Eryigit and L. Elfatimi, "Beans Leaf Diseases
Classification Using MobileNet Models," in IEEE Access, vol. 10, pp.
and increasing the quantity of the dataset. Addressing issues 9471-9482, 2022, doi: 10.1109/ACCESS.2022.3142817.
in vitamin deficiency identification, the system can better [14] Dr. Arati Dandavate, Priyanka Gore, Namita Naikwadi, Shrushti
detect vitamin deficiency with greater accuracy and Sable, Muskan Tilwani “Vitamin Deficiency Detection Using Image
efficiency. The research aspires to enhance the system's Processing and Artificial Intelligence”,April,2021
ability to provide personalized recommendations for https://www.irjet.net/archives/V8/i4/IRJET-V8I4642.pdf
addressing vitamin deficiencies, improving overall health [15] G. Sambasivam, J. Amudhavel and G. Sathya, "A Predictive
Performance Analysis of Vitamin D Deficiency Severity Using
outcomes, and reducing the time required for accurate Machine Learning Methods," in IEEE Access, vol. 8, pp. 109492-
diagnosis and treatment planning. 109507, 2020, doi: 10.1109/ACCESS.2020.3002191.
[16] Sagala, Noviyanti & Hayat, Cynthia & Tandipuang, Frahselia. (2020).
Identification of fat-soluble vitamins deficiency using artificial neural
REFERENCES network. Jurnal Teknologi dan Sistem Komputer. 8. 6-11.
10.14710/jtsiskom.8.1.2020.6-11.
[1] T. K. Chaithanya, J. Gupta and M. S. Roobini, "Vitamin Deficiency [17] A. S. Eldeen, M. AitGacem, S. Alghlayini, W. Shehieb and M. Mir,
Detection Using Neural Networks," 2024 International Conference on "Vitamin Deficiency Detection Using Image Processing and Neural
Wireless Communications Signal Processing and Networking Network," 2020 Advances in Science and Engineering Technology
(WiSPNET), Chennai, India, 2024, pp. 1-10, doi: International Conferences (ASET), Dubai, United Arab Emirates,
10.1109/WiSPNET61464.2024.10533044. 2020, pp. 1-5, doi: 10.1109/ASET48392.2020.9118303.
[2] M.Naga Keerthi, Kondapalli Bhargavi, “Vitamin Deficiency Detection [18] Hou, Wenjun & Perkowski, Marek. (2019). A Novel Machine Learning
Using Image Processing and Neural Network”,July,2024 Algorithm to Reduce Prediction Error and Accelerate Learning Curve
.http://www.jetir.org/papers/JETIR2407339.pdf for Very Large Datasets. 97-101. 10.1109/ISMVL.2019.00025.
[3] N. Srividhya, K. Divya, N. Sanjana, K. Krishna Kumari, M. [19] Z. Wu et al., "Studies on Different CNN Algorithms for Face Skin
Rambhupal “NOVEL METHOD VITAMIN DEFICIENCY EPRA Disease Classification Based on Clinical Images," in IEEE Access, vol.
International Journal of Multidisciplinary Research (IJMR) 2024, 10, 7, pp. 66505-66511, 2019, doi: 10.1109/ACCESS.2019.2918221.
1-6. doi:https://doi.org/10.36713/epra16299 [20] Khasoggi, Barlian & Ermatita, Ermatita & Samsuryadi, Samsuryadi.
[4] M. F. Ul Islam, M. Hasan, M. T. Rahman and A. Chakrabarty, (2019). Efficient mobilenet architecture as image recognition on
"Vitamin D Deficiency Detection: A Novel Ensemble Approach with mobile and embedded devices. Indonesian Journal of Electrical
Interpretability Insights," 2024 6th International Conference on Engineering and Computer Science. 16. 389.
Electrical Engineering and Information & Communication Technology 10.11591/ijeecs.v16.i1.pp389-394.
(ICEEICT), Dhaka, Bangladesh, 2024, pp. 137-142, doi: [21] C. -H. Weng, H. -Y. Chou, M. -H. Chen and Y. -S. Lin, "A portable
10.1109/ICEEICT62016.2024.10534371. platform for the quantification of vitamin D levels by using paper-
[5] Swapnil Bonde, Pravin Argade, Prof. Deepali M. Gohil, Rutik based microfluidic," 2018 11th Biomedical Engineering International
Gagare,Ganesh Wakade “ Vitamin Deficiency Detection Using Image Conference (BMEiCON), Chiang Mai, Thailand, 2018, pp. 1-3, doi:
Processing and DEEP-CNN Algorithm” 5,May,2024 .doi: 10.1109/BMEiCON.2018.8609988.
https://www.doi.org/10.56726/IRJMETS57869 [22] S. Yadav, A. Rathod, P. Patil and V. Hole, "Vitamins and Minerals
[6] Dr N. Durga Rao, K. Shanmuka Sivani Singh, I. Chethana, M. Uma Diagnosis System Using Neural Network," 2018 Second International
Madhuri, P. Madhuri “Vitamin Deficiency Detection using Conference on Inventive Communication and Computational
Convolution Neural Network with ADAM Optimization”,2024.doi: Technologies (ICICCT), Coimbatore, India, 2018, pp. 1921-1925, doi:
https://ijariie.com/AdminUploadPdf/VITAMIN_DEFICIENCY_DET 10.1109/ICICCT.2018.8473019.
ECTION_USING_CONVOLUTIONAL_NEURAL_NETWORK_W [23] Parmar, Ricky & Lakshmanan, Ramanathan. (2017). Detecting
ITH_ADAM_OPTIMIZATION_ijariie22832.pdf Diabetic Retinopathy from Retinal Images Using CUDA Deep Neural
[7] D. A. Abuhani, J. Khan and H. Sulieman, "Detecting Vitamin A Network. International Journal of Intelligent Engineering and Systems.
Deficiency in Schoolchildren Using an Enhanced Explainable Machine 10. 284-292. 10.22266/ijies2017.0831.30.
Learning Model," 2023 Advances in Science and Engineering [24] Ula, Mutammimul & Mursyidah, & Hendriana, Yana & Hardi, Richki.
Technology International Conferences (ASET), Dubai, United Arab (2016). An expert system for early diagnose of vitamins and minerals
Emirates, 2023, pp. 1-5, doi: 10.1109/ASET56582.2023.10180556. deficiency on the body. 1-6. 10.1109/ICITSI.2016.7858225.
[8] Parrott, Julie & Parrott, Austen & Rouhi, Armaun & Parrott, James & [25] Novaliendry, Dony & Yang, Cheng-Hong & Labukti, A.. (2015). The
Dumon, Kristoffel. (2023). What We Are Missing: Using Machine expert system application for diagnosing human vitamin deficiency
Learning Models to Predict Vitamin C Deficiency in Patients with through forward chaining method. 53-58.
Metabolic and Bariatric Surgery. Obesity Surgery. 33. 10.1109/ICTC.2015.7354493.
10.1007/s11695-023-06571-w.