Implementation of Deep Learning Methods to Identify Rotten Fruits

Full text

Implementation of Deep Learning Methods to
Identify Rotten Fruits
Sovon Chakraborty
Department of Computer Science and Engineering
European University of Bangladesh
Dhaka, Bangladesh
sovonchakraborty2014@gmail.com
Md. Masum Billah
Department of Software Engineering
Daffodil International University
Dhaka, Bangladesh
masum.swe.ndc@gmail.com
Md. Alauddin
Department of Computer Science and Engineering
European University of Bangladesh
Dhaka, Bangladesh
alauddin12340@gmail.com
F.M. Javed Mehedi Shamrat
Department of Software Engineering
Daffodil International University
Dhaka, Bangladesh
javedmehedicom@gmail.com
Md. Al Jubair
Department of Computer Science and Engineering
European University of Bangladesh
Dhaka, Bangladesh
jubair@eub.edu.bd
Rumesh Ranjan*
Department of Plant Breeding and Genetics
Punjab Agriculture University
Punjab, India
rumeshranjan@pau.edu
Abstract— Mostly in the agriculture sector, identifying rotten
fruits has been critical. The classification of fresh and rotting fruits
is typically carried out by humans, which is ineffective for fruit
growers. Humans wear out by doing the same role many days, but
robots do not. As a result, the study proposed a method for
reducing human effort, lowering production costs, and shortening
production time by detecting defects in agricultural fruits. If the
defects are not detected, the contaminated fruits can contaminate
the good fruits. As a result, we proposed a model to prevent the
propagation of rottenness. From the input fruit images, the
proposed model classifies the fresh and rotting fruits. We utilized
three different varieties of fruits in this project: apple, banana, and
oranges. The features from input fruit images are collected using
a Convolutional Neural Network, and the images are categorized
using Max pooling, Average pooling, and MobileNetV2
architecture. The proposed model's performance is tested on a
Kaggle dataset, and it achieves the highest accuracy in training
data is 99.46% and in the validation set is 99.61% by applying
MobileNetV2.The Max pooling achieved 94.49% training
accuracy and validation accuracy is 94.97%. Besides, the Average
pooling achieved 93.06% training accuracy and validation
accuracy is 93.72%. The findings revealed that the proposed CNN
model is capable of distinguishing between fresh and rotting fruits.
Keywords—rotten fruit detection; CNN; max-pooling; average
pooling; MobileNetV2; deep learning
I. INTRODUCTION
Computer vision approaches have improved the efficiency
of image classification tasks, particularly in the fields of
machine learning [1] and deep learning [2-6]. One of the main
problems in the agricultural fields is the detection of defective
fruit and the identification of new and rotten fruits. If not
correctly classified and can also impact productivity, rotten
fruits can cause harm to other fresh fruits. This designation is
traditionally performed by hard-working men, time-consuming
and not effective. Moreover, manufacturing costs are often
increased. We also need an integrated system that reduces
human efforts, increases productivity, and reduces production
costs and production time.
In the paper [7], a CNN model is proposed for feature
extraction from an input image of fruits that are apple, banana,
and orange. For classification, a Softmax classifier is used on the
images. To compare the accuracy with the proposed model,
VGG16, VGG19, Xception, and MobileNet transfer learning
models are used which shows that the proposed model exceeds
in accuracy. K. Roy et al. [8] proposed a method that implements
the segmentation technique to detect rotten fruits. Marker-based
segmentation, color-based segmentation, and edge detection
techniques are utilized after the image data is converted to
greyscale, and filtering and thresholding to reduce noise. In the
final output, rotten fruit is detected and marked. The authors in
the paper [9] proposed a semantic segmentation technique using
uNet and En-UNet deep learning architecture to detect rotting in
fruit from image data. Before training the data, it is converted to
greyscale from the raw RGB image and later masked by using
thresholding and inverse binarization. Finally, the obtained
masked binary image is trained using the deep learning [10-11]
methods. The objective of the paper [12] is to propose a method
that uses a segmentation method to detect rotten or fresh fruits.
The image of the fruits is rectified by detecting the foreground
using ‘YCbCr’ color space.to segment out the essential portion
5th International Conference on Trends in Electronics and Informatics (ICOEI 2021)
Tirunelveli, India, 3-5, June 2021
Pre-Print

of the image, ‘L*a*b*’ color space and KNN clustering method
is used. Finally, to identify the rotten portion, segmentation is
done using a color map.
The rest of the paper is in the same arrangement. The most
rapid current developments in rotten fruits identification are
discussed in this sector. Section II describes the analytical
methodology for the construction of the whole system. The
result of the structure produced is examined in Section III.
Section IV finishes with observation and deficiencies and plans
for potential work.
II. RESEARCH METHODOLOGY
Bangladesh's agriculture sector is most significant. The
agriculture sector of Bangladesh accounts for 14.2% of the GDP
of Bangladesh, providing 42.7% of working countries with
employment. It is necessary to eliminate the possibility of
foodborne disease in order to improve the average longevity of
human beings. People in a risky community depend mostly on
fruit and vegetables. It is therefore essential to distinguish rotting
or fruits from healthy ones in order to ensure their protection.
Automation technology is an integral part of life nowadays.
Bangladesh is a nation dependent on agri-based farming.
Agriculture is their principal source of wealth. The selling is
widening every day of fresh fruit. Health-conscious people
choose only healthy raw fruits of quality.
The 21st century is seeing an increasingly dynamic role in
the fruit and food manufacturing sectors [13]. Global exchange
and fruit and vegetable demand flow decide the proximity
between exporters and importers. For the exportation or
importation of rotten or almost rotten fruit, there is a long and
time-consumed transportation method that impedes quality
control of a vast number of fruits. As a result, fruit output is
expected to fall more compared with the world fruit production
and trade of previous years. Other main causes of concern
behind the decline in commerce are not just all other challenges,
but also volatile environment trends, climate change, and
temperature growth. Besides, the food industry has been
seriously impaired, aside from the export and importation of
fresh fruits, due to the monitoring of the nature of the rotten fruit.
Fig. 1. Proposed System diagram.
A. Dataset Collection:
We used a dataset from kaggle.com for this study. At first,
the dataset is fresh fruits and rotten fruits for classification
(https://www.kaggle.com/sriramr/fruits-fresh-and-rotten-for-
classification). The data set is divided into 6 categories, as
follows:
• Fresh Apples
• Fresh Oranges
• Fresh Bananas
• Rotten Apples
• Rotten Oranges
• Rotten Bananas
The dataset contains 13599 images that were used for
validation and training.

Fig. 2. Datasets Images Sample.
B. Preprocessing and augmentation of Data :
The images in the dataset are not all the same size, so
preprocessing was needed for this study. Deep learning [14-16]
models require a significant amount of data for training rather
than machine learning [17-21]. We used Keras'
ImageDataGenerator tool to resize all of the images to 256 x 256
pixels. We normalized both images after transforming them to
256 X 256. For faster calculation, images are converted to
NumPy arrays. The volume of data may be increased by rotating,
zooming, shearing, and flipping horizontally. Photos are
obtained as well. The photos are then reshaped into 128 x 128
pixels for passing into the second convolution layer, and then
down to 64 x 64 pixels for passing into the third convolution
layer.
C. Proposed Convolution Neural Network (CNN) architecture
For classification and image recognition, CNN is used. One
or two convolution layers compose a CNN. Rather than dealing
with the entire picture, CNN tries to identify elements that are
useful inside it. There are several hidden layers in CNN, as well
as an input layer and an output layer. In this study, we used a
deep CNN with three convolution layers. Convolution is a
technique for merging two mathematical functions to create a
single one. Our CNN model's working process is depicted in Fig.
3.
Fig. 3. Three Convolution Layer with Max pooling operation.
Again, the same architecture is applied with average pooling
operation for feature mapping this time. Fig. 4 demonstrates the
working procedure of the model. Max pooling takes the highest
number inside the region of interest of the image matrix where
Average pooling takes the average of all values of that region.
Our CNN model initiates with Keras.models.sequential(). Relu
activation function is applied in the first hidden layer then Max
pooling operation is applied. Max pooling helps to gather
significant information and reduces the size of the images. Then
the data is passed to the second convolution layer. For getting
the most notable information max pooling is applied again. The
obtained image matrix is then flattened and trained. For
observing the performance of the model, we trained our model
with the Average pooling operation instead of the max pooling
operation. Adam stochastic gradient descent algorithms have
been used for training with better accuracy. For training
purposes, we use 80% images of our dataset.
Fig. 4. Three convolution layers with Average Pooling operation.
D. MobileNetV2 Architecture
MobileNetV2 is extremely effective for image classification.
MobileNetV2 is a lightweight deep learning model built on the
CNN that provides the weight of the image through TensorFlow.
The base layer is first stripped and a new trainable layer is
applied to MobileNetV2. The model operates on the data
collection obtained and defines the most correlated features of
our images. MobileNetV2 is consisting of 19 layers of
bottleneck [22]. OpenCV, which uses ResNet-10 in the base
model [22], was included. Caffemodel from OpenCV is used to
detect the front side of a fruit image. Then it extracts the
knowledge needed and transmits it to the fruit classifier layer
Overfitting in machine learning is a significant concern. For
ignoring our model to be overfitted with the dataset we have
used the Dropout layer. With MobileNetV2 (include top=False)
we removed the base layer. The photos have been reshaped. Our
model contains 256 hidden layers and is implemented with a
pool size average pooling operation (7,7). Relu activation
function is applied in the hidden layer and softmax activation
function in the fully connected layer. Relu activation function is
applied in the hidden layer and softmax activation function in
the fully connected layer. We define a learning rate of 0.001 for
better accuracy. Adam's stochastic gradient descent algorithm
helps the model for a better understanding of image features.
MobileNetV2 working layer depicted in Fig. 5.

Fig. 5. MobileNetV2 Architecture.
E. Evaluating performance using performance matrix:
After completing the training and testing phase, we have
measured the performance of two models using precision, recall,
f1-score, and accuracy. We have used the following formula's,
  
 (1)
  
 (2)
  
 (3)
    
 (4)
III. EXPERIMENTAL RESULT ANALYSIS:
For detecting fresh and rotten fruits from images we have
used a dataset consist of 13599 images. Table I describes the
training accuracy and validation accuracy after applying the
Deep CNN model where Max Pooling is applied to reduce the
dimension of our image feature map. The highest accuracy in
training data is 94.49% and in the validation set is 94.97%.
TABLE I. OUTCOMES FOR DEEP CNN AFTER APPLYING MAX POOLING
OF DIFFERENT EPOCHS
Epoch
Training
Loss
Training
Accuracy
Validation
Loss
Validation
Accuracy
1
47.13%
87.36%
15.37%
89.99%
2
12.01%
88.17%
10.13%
90.32%
3
9.45%
88.39%
9.01%
91.04%
4
8.81%
90.58%
6.88%
91.70%
5
8.04%
90.65%
6.55%
91.83%
6
7.40%
91.04%
6.16%
92.33%
7
7.33%
91.12%
5.98%
92.83%
8
6.91%
91.95%
5.77%
93.26%
9
6.76%
93.01%
5.14%
93.72%
10
6.23%
93.87%
5.04%
94.08%
11
5.97%
94.04%
4.84%
94.36%
12
5.89%
94.48%
4.67%
94.71%
13
5.84%
94.49%
4.12%
94.97%
Fig. 6 shows the training accuracy and validation accuracy
graph. The same CNN architecture is applied later where
Average Pooling is used to reduce the dimensions of feature
map. The predicted result shows less accuracy than the previous
model. Table II shows the predicted outcomes where maximum
training accuracy is 93.06% with a training loss of 6.96% and
the validation accuracy is 93.72%.
Fig. 6. Test Accuracy and Training Accuracy for CNN with Max Pooling
Layer.
TABLE II. OUTCOMES FOR DEEP CNN AFTER APPLYING AVERAGE
POOLING OF DIFFERENT EPOCHS
Epoch
Training
Loss
Training
Accuracy
Validation
Loss
Validation
Accuracy
1
44.12%
86.89%
11.12%
88.23%
2
12.65%
87.12%
10.76%
88.87%
3
12.23%
87.46%
8.24%
89.41%
4
11.42%
89.02%
8.14%
89.95%
5
11.05%
89.26%
8.04%
90.12%
6
10.24%
89.95%
7.64%
91.23%
7
9.65%
90.00%
7.34%
91.42%
8
8.98%
91.14%
7.16%
92.77%
9
8.97%
91.14%
7.08%
92.83%
10
8.76%
91.24%
7.03%
92.91%
11
7.25%
92.54%
6.53%
93.13%
12
7.10%
92.88%
6.34%
93.62%
13
6.96%
93.06%
5.28%
93.72%
Fig. 7 shows the graph of relative validation accuracy and
training accuracy for each epoch.
Fig. 7. Test Accuracy and Training Accuracy for CNN with Average Pooling
Layer.
After applying MobileNetV2 architecture the accuracy
improved significantly. Table III describes the validation and
test accuracy concerning each epoch.

TABLE III. DIFFERENT OUTCOMES AFTER APPLYING MOBILENETV2
ARCHITECTURE
Epoch
Training
Loss
Training
Accuracy
Validation
Loss
Validation
Accuracy
1
4.53%
98.02%
4.21%
98.75%
2
4.36%
98.03%
4.12%
98.81%
3
4.32%
98.14%
4.09%
98.92%
4
4.26%
98.24%
3.89%
99.01%
5
3.23%
98.52%
3.72%
99.05%
6
3.96%
98.59%
3.61%
99.12%
7
3.46%
98.71%
3.56%
99.42%
8
3.43%
98.92%
3.41%
9946%
9
3.25%
98.98%
3.23%
99.52%
10
3.24%
99.12%
3.20%
99.55%
11
3.22%
99.43%
3.18%
99.57%
12
3.22%
99.45%
3.18%
99.58%
13
3.15%
99.46%
3.18%
99.61%
From Table III, the highest accuracy is achieved at 99.46%
for validation data and 99.61% for training data. The data loss in
the validation phase is only 3.15%. Fig. 8 shows the detailed
comparison of test accuracy and validation accuracy of
MobilenetV2 which is a CNN-based architecture. We have also
calculated the confusion matrix after applying MobilenetV2
architecture. Table IV describes the confusion matrix properly.
TABLE IV. CONFUSION MATRIX AFTER APPLYING MOBILENETV2
Class
Precision
Recall
F1 - Score
0 [Fresh Apples]
98%
99%
97%
1 [Fresh Oranges]
99%
99%
99%
2 [Fresh Bananas]
99%
98%
97%
3 [Rotten Apples]
98%
99%
98%
4 [Rotten Oranges]
99%
99%
99%
5 [Rotten Bananas]
99%
98%
98%
Fig. 8. Test Accuracy and Training Accuracy for MobilenetV2 with Average
Pooling Layer
The MobilenetV2 design outperformed many of the other
models included in this study. This model is capable of
recognizing the mask in a picture. In Fig. 9 and 10 showing the
detection result of MobileNetV2.
Fig. 9. Detection of fresh apples from dataset images.
Fig. 10. Detection of rotten apples from dataset images.
The Max pooling achieved 94.49% training accuracy and
validation accuracy is 94.97%. Besides, the Average pooling
achieved 93.06% training accuracy and validation accuracy is
93.72%. MobileNetV2 architecture gained the highest accuracy
99.46% for training and 99.61% for validation. A short
explanation is added in Table V.
TABLE V. COMPARISON WITHIN THE CNN TECHNIQUES
Epoch
Training
Loss
Training
Accuracy
Validation
Loss
Validation
Accuracy
Max Pooling
13
5.84%
94.49%
4.12%
94.97%
Average
Pooling
13
6.96%
93.06%
5.28%
93.72%
MobileNetV2
13
3.15%
99.46%
3.18%
99.61%
IV. CONCLUSION AND FUTURE WORK
In the fruit processing industry, computer vision has a broad
variety of uses, enabling processes to be automated. For the
industry manufacturing unit to produce the highest quality
finished food products and the finest quality raw fruits to be able
to be sold in the sector, classification of fruit quality and thus
grading of the same is very necessary. In this study, we used two
deep CNN architectures and one CNN-based MobilenetV2
architecture in this study. Our main goal was to propose a
suitable model with high accuracy such that fruit detection could
be simplified in the agricultural sector. In order to assess
performance with a wider dataset, we can attempt to add further

models to compare with Mobilenetv2. In the future, we will
integrate this model with IoT [23-27] to detect rotten fruits
automatically by AI and IoT.
REFERENCES
[1] P. Ghosh, S. Azam, A. Karim, M. Jonkman, MDZ Hasan, “Use of
Efficient Machine Learning Techniques in the Identification of Patients
with Heart Diseases,” 5th ACM International Conference on Information
System and Data Mining (ICISDM2021), 2021.
[2] P. Ghosh et al., "Efficient Prediction of Cardiovascular Disease Using
Machine Learning Algorithms With Relief and LASSO Feature Selection
Techniques," in IEEE Access, vol. 9, pp. 19304-19326, 2021, doi:
10.1109/ACCESS.2021.3053759.
[3] F. M. Javed Mehedi Shamrat, Z. Tasnim, P. Ghosh, A. Majumder and M.
Z. Hasan, "Personalization of Job Circular Announcement to Applicants
Using Decision Tree Classification Algorithm," 2020 IEEE International
Conference for Innovation in Technology (INOCON), Bangluru, India,
2020, pp. 1-5, doi: 10.1109/INOCON50539.2020.9298253.
[4] M. S. Junayed, A. A. Jeny, S. T. Atik, N. Neehal, A. Karim, S. Azam, and
B. Shanmugam, “AcneNet - A Deep CNN Based Classification Approach
for Acne Classes,” 2019 12th International Conference on Information &
Communication Technology and System (ICTS), 2019.
[5] F. M. Javed Mehedi Shamrat, P. Ghosh, M. H. Sadek, M. A. Kazi and S.
Shultana, "Implementation of Machine Learning Algorithms to Detect the
Prognosis Rate of Kidney Disease," 2020 IEEE International Conference
for Innovation in Technology (INOCON), Bangluru, India, 2020, pp. 1-7,
doi: 10.1109/INOCON50539.2020.9298026.
[6] P. Ghosh, F. M. Javed Mehedi Shamrat, S. Shultana, S. Afrin, A. A.
Anjum and A. A. Khan, "Optimization of Prediction Method of Chronic
Kidney Disease Using Machine Learning Algorithm," 2020 15th
International Joint Symposium on Artificial Intelligence and Natural
Language Processing (iSAI-NLP), Bangkok, Thailand, 2020, pp. 1-6, doi:
10.1109/iSAI-NLP51646.2020.9376787.
[7] Palakodati, S.S.S., Chirra, V.R., Dasari, Y., Bulla, S. (2020). Fresh and
rotten fruits classification using CNN and transfer learning. Revue
d'Intelligence Artificielle, Vol. 34, No. 5, pp. 617-622.
https://doi.org/10.18280/ria.340512
[8] K. Roy, S. S. Chaudhuri, S. Bhattacharjee, S. Manna and T. Chakraborty,
"Segmentation Techniques for Rotten Fruit detection," 2019 International
Conference on Opto-Electronics and Applied Optics (Optronix), Kolkata,
India, 2019, pp. 1-4, doi: 10.1109/OPTRONIX.2019.8862367.
[9] Roy, K., Chaudhuri, S.S. & Pramanik, S. Deep learning based real-time
Industrial framework for rotten and fresh fruit detection using semantic
segmentation. Microsyst Technol (2020). https://doi.org/10.1007/s00542-
020-05123-x
[10] Shakya, Subarna, Lalitpur Nepal Pulchowk, and S. Smys. Anomalies
Detection in Fog Computing Architectures Using Deep Learning. Journal:
Journal of Trends in Computer Science and Smart Technology March
2020, no. 1 (2020): 46-55.
[11] Suma, V. A Novel Information retrieval system for distributed cloud
using Hybrid Deep Fuzzy Hashing Algorithm. JITDW 2, no. 03 (2020):
151-160.
[12] K. Roy, A. Ghosh, D. Saha, J. Chatterjee, S. Sarkar and S. S. Chaudhuri,
"Masking based Segmentation of Rotten Fruits," 2019 International
Conference on Opto-Electronics and Applied Optics (Optronix), Kolkata,
India, 2019, pp. 1-4, doi: 10.1109/OPTRONIX.2019.8862396.
[13] F. M. Javed Mehedi Shamrat, Md Asaduzzaman, Pronab Ghosh, Md Dipu
Sultan, and Zarrin Tasnim, “A Web Based Application for Agriculture:
“Smart Farming System”” International Journal of Emerging Trends in
Engineering Research, Volume 8, Issue 06, June 2020, pp: 2309-2320,
ISSN: 2347- 3983. DOI: https://doi.org/10.30534/ijeter/2020/18862020.
[14] M. F. Foysal, M. S. Islam, A. Karim, and N. Neehal, “Shot-Net: A
Convolutional Neural Network for Classifying Different Cricket Shots,”
Communications in Computer and Information Science, pp. 111–120,
2019.
[15] F.M. Javed Mehedi Shamrat, Md. Asaduzzaman, A.K.M. Sazzadur
Rahman, Raja Tariqul Hasan Tusher, Zarrin Tasnim “A Comparative
Analysis of Parkinson Disease Prediction Using Machine Learning
Approaches” International Journal of Scientific & Technology Research,
Volume 8, Issue 11, November 2019, ISSN: 2277-8616, pp: 2576-2580.
[16] Junayed M.S., Jeny A.A., Neehal N., Atik S.T., Hossain S.A. (2019) A
Comparative Study of Different CNN Models in City Detection Using
Landmark Images. In: Santosh K., Hegadi R. (eds) Recent Trends in
Image Processing and Pattern Recognition. RTIP2R 2018.
Communications in Computer and Information Science, vol 1035.
Springer, Singapore. https://doi.org/10.1007/978-981-13-9181-1_48.
[17] A.K.M Sazzadur Rahman, F. M. Javed Mehedi Shamrat, Zarrin Tasnim,
Joy Roy, Syed Akhter Hossain “A Comparative Study on Liver Disease
Prediction Using Supervised Machine Learning Algorithms”
International Journal of Scientific & Technology Research, Volume 8,
Issue 11, November 2019, ISSN: 2277-8616, pp: 419-422.
[18] F. M. Javed Mehedi Shamrat, Md. Abu Raihan, A.K.M. Sazzadur
Rahman, Imran Mahmud, Rozina Akter, “An Analysis on Breast Disease
Prediction Using Machine Learning Approaches” International Journal of
Scientific & Technology Research, Volume 9, Issue 02, February 2020,
ISSN: 2277-8616, pp: 2450-2455.
[19] K. Mahmud, S. Azam, A. Karim, S. Zobaed, B. Shanmugam, and D.
Mathur, “Machine Learning Based PV Power Generation Forecasting in
Alice Springs,” IEEE Access, pp. 1–1, 2021.
[20] F. M. Javed Mehedi Shamrat, Zarrin Tasnim, Imran Mahmud, Ms. Nusrat
Jahan, Naimul Islam Nobel, “Application Of K-Means Clustering
Algorithm To Determine The Density Of Demand Of Different Kinds Of
Jobs”, International Journal of Scientific & Technology Research,
Volume 9, Issue 02, February 2020, ISSN: 2277-8616, pp: 2550-2557.
[21] F. M. Javed Mehedi Shamrat, Pronab Ghosh, Md. Dipu Sultan, Anup
Majumder and Imran Mahmud,“Software-Defined Networking with
Multipath Routing Utilizing DFS to Improving QoS” 2nd International
Conference on Emerging Technologiesin Data Mining and Information
Security (IEMIS 2020).
[22] An automated System to limit covid 19 using facial mask detection in
smart city network( 2020, IEEE)
https://ieeexplore.ieee.org/document/9216386.
[23] Javed Mehedi Shamrat F.M., Allayear S.M., Alam M.F., Jabiullah M.I.,
Ahmed R. (2019) A Smart Embedded System Model for the AC
Automation with Temperature Prediction. In: Singh M., Gupta P., Tyagi
V., Flusser J., Ören T., Kashyap R. (eds) Advances in Computing and
Data Sciences. ICACDS 2019. Communications in Computer and
Information Science, vol 1046. Springer, Singapore.
https://doi.org/10.1007/978-981-13-9942-8_33.
[24] A. Karim, S. Azam, B. Shanmugam, and K. Kannoorpatti, “Efficient
Clustering of Emails Into Spam and Ham: The Foundational Study of a
Comprehensive Unsupervised Framework,” IEEE Access, vol. 8, pp.
154759–154788, 2020.
[25] F. M. Javed Mehedi Shamrat, Zarrin Tasnim, Naimul Islam Nobel, and
Md. Razu Ahmed. 2019. An Automated Embedded Detection and Alarm
System for Preventing Accidents of Passengers Vessel due to Overweight.
In Proceedings of the 4th International Conference on Big Data and
Internet of Things (BDIoT'19). Association for Computing Machinery,
New York, NY, USA, Article 35, 1–5.
DOI:https://doi.org/10.1145/3372938.3372973
[26] Shamrat F.M.J.M., Nobel N.I., Tasnim Z., Ahmed R. (2020)
Implementation of a Smart Embedded System for Passenger Vessel
Safety. In: Saha A., Kar N., Deb S. (eds) Advances in Computational
Intelligence, Security and Internet of Things. ICCISIoT 2019.
Communications in Computer and Information Science, vol 1192.
Springer, Singapore. https://doi.org/10.1007/978-981-15-3666-3_29
[27] F.M. Javed Mehedi Shamrat, Shaikh Muhammad Allayear and Md. Ismail
Jabiullah "Implementation of a Smart AC Automation System with Room
Temperature Prediction ", Journal of the Bangladesh Electronic Society,
Volume 18, Issue 1-2, June-December 2018, ISSN: 1816-1510, pp: 23-32