Conference PaperPDF Available

Domestic Trash Classification with Transfer Learning Using VGG16



Content may be subject to copyright.
Domestic Trash Classification with Transfer
Learning Using VGG16
Haruna Abdu
School of Computer Sciences
Universiti Sains Malaysia
Pulau Pinang, Malaysia
Mohd Halim Mohd Noor
School of Computer Sciences
Universiti Sains Malaysia
Pulau Pinang, Malaysia
Abstract—Environmental contamination is a major issue
affecting all inhabitants living in any environment. The domestic
environment is engulfed with many trash items such as solid and
toxic trashes, leading to severe environmental contamination
and causing life-threatening diseases if not appropriately
managed. Trash classification is at the heart of these issues
because the inability to classify the trash leads to difficulty in
recycling. Humans categorize trash based on what they
understand about the trash object rather than on
the recyclability status of an object, which frequently leads to
incorrect classification in manual classification. Additionally,
coming into contact with toxic waste directly could be physically
dangerous for those involved. Few machine learning and Deep
Learning (DL) techniques were proposed using benchmarked
trash classification datasets. However, most benchmarked
datasets used to train DL models have a transparent or white
background, which leads to a lack of model generalization,
particularly in the real world. In this paper, we propose a Deep
Learning model based on the VGG16 Architecture that can
accurately classify various types of trash objects. On the
TrashNet dataset plus the images collected in the wild, we
achieved an accuracy of more than 96%.
Keywords—Trash, classification, Transfer Learning,
Domestic Trash, Deep Learning
A waste management system (WMS) is a company's plan
for getting rid of, reducing, reusing, and preventing waste.
incineration, Recycling, landfills, composting,
bioremediation, waste to energy conversion, and waste
minimization, etc. are all methods of waste disposal. The
production of waste has dramatically increased in recent years.
According to World Bank data, global solid waste generation
in 2016 was approximately 2.01 billion tonnes per year. The
world is expected to produce 2.01 and 3.40 billion tonnes by
2030 and 2050, respectively [1], [2]. Failure to manage trash
can have disastrous consequences for almost any
environment. Because there is so much waste, waste detection
and sorting should be done early in the waste management
process to increase the number of items that can be recycled
while simultaneously decreasing the likelihood that other
items will pollute the environment.
The increasing amount of trash in all environments
endangers human and animal life. Poorly managed and openly
deposited trash harms the environment, threatens the health of
local residents, causes water and air pollution,
contamination/degradation of land, and a variety of other
consequences [3]. Illegal trash burying occurs in areas not
technically designated as dump sites for toxic waste, such as
land suitable for cultivation, highways, buildings, and
construction sites, as well as occasionally inside homes or
Because of the difficulties posed by improper
garbage/trash depositions in undesignated locations [4] many
people have used various techniques to detect and classify
trash. However, most trash classification methods rely on
human expertise, which can sometimes be difficult and time-
consuming. With further subdivision in a different
environment, trash can be hazardous or non-hazardous. Some
specific features considered in garbage classification are the
physical state, technical elements, reusable potentials,
biodegradable potential, manufacturing source, and
environmental effects [5].
A lot of research has been conducted by different
researchers in relation to waste management. There is some
research that focuses on trash classification. The classification
of medicinal waste is the goal of the EnSegNet-CFE-DNN-
TC system, which is described in [6]. This is accomplished by
resolving misjudgment problems within a deep learner
classifier. The EnSegNet model began by segmenting the
medicinal trash images that were fed into it. The segmented
trash images are used to generate a number of different texture
features, such as GLCM., MLBP, LDP, and LTP, as well as
deep features, are extracted, achieving an accuracy of 87.5%.
TrashQNet is a Classical-Quantum Transfer learning model
that was proposed in [7] for the purpose of classifying trash
into two distinct categories: organic trash and recyclable trash.
A previously trained DenseNet169 network and a variational
quantum circuit serve as the feature extractor and classifier,
respectively, in the TrashQNet model that was proposed. The
performance of TrashQNet is compared to that of traditional
machine learning models such as K-Nearest Neighbor and
Support Vector Machine, as well as models of deep learning
and transfer learning such as Convolutional Neural Networks..
According to the authors, TrashQNet outperforms all of these
models, achieving an accuracy of 94% on a test dataset. In yet
another piece of research, author [8] proposes an innovative
approach to the detection and classification of waste in order
to address issues related to waste management. A number of
different kinds of deep neural networks are brought together
in this detection method., to classify, and quantify waste size.
with a 90% accuracy. The trash, on the other hand, is
unclassified. Some other research attempt to develop models
that classify trash are [9], [10]. However, none of this research
focuses on classifying trash images in the wild. Only a few
datasets exist for the images in the wild such as [11].
The primary goal of this research study is to encourage
researchers to use DL techniques to solve various waste
management (WM) problems, such as waste detection,
classification, and prediction. We aimed to classify trash
objects in the wild using a modified VGG16 model based on
a dataset collected by hand and some images from the trash
net dataset. In this research, more than half the trash net
dataset was used in combination with manually collected trash
images in the wild, that are gotten from the internet and most
2022 IEEE 12th International Conference on Control System, Computing and Engineering (ICCSCE), 21–22 October 2022, Penang, Malaysia
978-1-6654-8339-1/22/$31.00 ©2022 IEEE
2022 IEEE 12th International Conference on Control System, Computing and Engineering (ICCSCE) | 978-1-6654-8339-1/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICCSCE54767.2022.9935653
Authorized licensed use limited to: UNIVERSITI TEKNOLOGI MARA. Downloaded on November 08,2022 at 07:08:42 UTC from IEEE Xplore. Restrictions apply.
from the field. A new model is proposed based on the VGG16
model architecture and used for trash classification.
Countless academics have begun researching this area to
promote waste sorting and recycling. For trash image
classification, [12] employs the Trashnet dataset, which
contains six classes of trash objects. Support vector machines
(SVM) with scale-invariant feature transform (SIFT) features
and a convolutional neural network were among the models
that were utilised (Convolutional Neural Networks (CNN)).
The SVM outperformed the CNN in their experiment;
however, the CNN was not trained to its full potential due to
difficulties in determining optimal hyperparameters. The
Support Vector Machine (SVM) was superior to the Neural
Network in terms of performance. It achieved 63 percent using
a split of the data for the train and the test that was 70/30. A
neural network that had a train/test split of 70/30 was able to
achieve an accuracy of 27% when being tested.
[13] developed the model known as RecycleNet, which is
an architecture for a deep convolutional neural network that
has been carefully optimized for the classification of specific
object classes that can be recycled.. The Trashnet dataset was
also used, and many deep learning models were tested for
waste classification using both saved model weight and
training from scratch..
To improve the accuracy of object classification models,
Deep Learning models can be hybridized. [14] Uses 5000
images with a resolution of 640 x 480 pixels and a plain grey
background. When the objects being investigated have strong
image features, the Multilayer Hybrid System as well as the
CNN perform exceptionally well. When waste items, such as
"other" waste, lack distinguishing image features, CNN
performs poorly. In two distinct testing scenarios, MHS
achieves significantly higher classification performance than
the reference model does: the overall performance accuracies
are 98.2 and 91.6 %, respectively (the accuracy of the
reference model is 87.7 and 80.0 %).. The item is oriented in
both fixed and random directions.
[15] propose a deep learning approach for medical waste
identification and classification because trash can belong to
different environments. The authors propose Res-NeXt, a
deep learning-based classification method that successfully
identified 8 types of medical waste with an accuracy of 97.2
% on 3480 images; the average F1-score of five-fold cross-
validation was 97.2 %.
[16] proposed the DSCR-Net model. The study generates
a large sample size open-source dataset. It is the first open-
source dataset to use this method of classification, which is
based on the regulations regarding the management of
household waste that were established by the Shanghai
Municipal Government. In order to make migration easier, the
new algorithm for the study has taken inspiration from
Inception-V4 and the ResNet network. Additionally, some of
the model's layers have been modified. After undergoing
optimization and testing on the dataset, the new algorithm
achieved an accuracy rate of 94.38 %.
The majority of trash classification models concentrate on
a single object in an image. [17] made an effort to recognise
and label a single piece of trash contained within an image.
Support vector machines (SVM) with HOG features, simple
convolutional neural networks (CNN), and CNN with residual
blocks are the models that are utilised. The findings of the
evaluation indicate that straightforward CNN networks, with
or without residual blocks, perform admirably. [18]
investigated a single trash class in addition to single object
Different types of waste require different management
techniques; thus, proper waste segregation by type is required
to facilitate proper recycling. Manual hand-picking is still
used in the current method of segregation. In light of concepts
from both deep learning and computer vision, the paper [19]
contains a description of a process that uses images to divide
waste into six distinct categories (glass, metal, paper, plastic,
cardboard, and others). For the purpose of waste classification,
a multiple-layered Convolutional Neural Network (CNN)
model was utilised. More specifically, the well-known
Inception-v3 model was utilised, and a trained dataset was
obtained from online sources. The proposed method achieves
92.5 % classification accuracy. Much more research is being
conducted on trash image classification from various
environments using various benchmark datasets. Some of
these studies include[9], [10], [20]–[27].
In summary, in all the reviewed research literature none of
them attempted to consider classifying trash images in the
wild to train a DL model for the trash classification, as most
of the datasets used with a clear background will not make the
model generalize especially in a realtime or real-world
application for trash detection, segregation, and classification.
This paper attempted to address such an issue by manually
collecting trash images in the wild.
This section analyzes the neural network pipelines and the
dataset that was used in this research study. The VGG16
model was modified in this study to create the new trash
model. The images that are read in are then sent to the
preprocessing network, which performs operations such as
cropping, normalization, flipping, and subsetting. After that,
the preprocessed images are fed into a modified Very Deep
Convolutional Neural Network (VGG16) [28]. The model
developed for this research is based on the VGG16
architecture (as shown in figure 1) with some key features;
The Visual Geometry Group from Oxford is the inspiration
behind another name for the VGG16 model, which is the
OxfordNet model. The number 16 indicates that there are a
total of 16 layers, each of which carries a certain weight. The
only types of layers that are contained in it are multiple
convolutional and pooling layers. In each layer it always uses
a 3 × 3 Kernel for convolution and 2 × 2 size of the max
pool. The model also has a total of about 138 million trainable
parameters. The data from TrashNet were initially used to
train and validate the model, which resulted in an accuracy of
92.7% being achieved.
FIGURE 1: VGG-16 Architecture
2022 IEEE 12th International Conference on Control System, Computing and Engineering (ICCSCE), 21–22 October 2022, Penang, Malaysia
Authorized licensed use limited to: UNIVERSITI TEKNOLOGI MARA. Downloaded on November 08,2022 at 07:08:42 UTC from IEEE Xplore. Restrictions apply.
Generally, a complete explanation of the VGG16
architecture is as follows: The network's input is a
dimensioned image (224, 224, 3). The first two layers have 64
channels with the same padding and a 3*3 filter size.
Following a max pool layer of stride (2, 2), two layers with
convolution layers of 128 filter size and filter size are added
(3, 3). This is followed by a stride (2, 2) max-pooling layer
that is the same as the previous layer. There are however two
convolution layers with filter sizes of (3, 3) and 256 filters.
Following that, there are two sets of three convolution layers
and a max pool layer. Each has 512 filters of the same size (3,
3) with the same padding. This image is then fed into a
convolution layer stack of two. A 1-pixel padding (the same
padding) is provided after each convolution layer to prevent
the spatial feature of the image. We obtained a (7, 7, 512)
feature map after adding a convolution and max-pooling layer
to the stack. This output is flattened to create a (1, 25088)
feature vector. There are then 3 fully connected layers; the first
layer uses the most recent feature vector as input and produces
a vector of size (1, 4096); the second layer also produces a
vector of size (1, 4096); however, the third layer produces a
vector of size (1, 1000), which is used to implement the
softmax function to classify 1000 classes. ReLU is used by
every hidden layer as its activation function. Because ReLU
promotes quicker learning and lessens the likelihood of
vanishing gradient issues, it is more computationally efficient.
A. The Dataset
The trash net dataset is the most used benchmark dataset
for trash classification [29]. The trash net dataset includes six
categories: waste, glass, paper, cardboard, plastic, and metal.
The dataset comprises 2,527 photos labelled with a category
(501 glass, 594 paper, 403 cardboard, 482 plastic, 410 metal,
and 137 rubbish/trash). The data set is comprised of
photographs of trash that were taken against a white
background using a variety of exposure and lighting settings
(mainly one object per photo). The trash net dataset poses a
challenge of serious misclassification and/or none-
generalization of any machine learning or deep learning
model, mainly when applied to a real-life scenario. In this
research, we use 2,650 trash images in total, where almost half
of the photos are from the trash net dataset. The remaining
were manually collected from the web, and most were from
around the university environment using a smartphone. The
manually collected photos are the images in the wild
corresponding to the equivalent classes in the trash net dataset
(cardboard, glass, metal, paper, plastic, and trash). Figure 2
and Figure 3 show the image samples used from the trash net
dataset and the collected images in the wild, respectively.
1. Metal 2. Glass 3. Paper
4. Trash 5. Plastic 6. Cardboard
FIGURE 2: Trash Net Image Sample
1. Metal 2. Glass 3. Paper
4. Trash 5. Plastic 6.Cardboard
FIGURE 3: Images Collected from the wild Samples
In addition, all the classes contain 450 images from within the
total images used in this research, except for the cardboard
class, which has only 400 images. The total images within
each class are further split into training, validation and testing
at 300, 100, and 50, respectively.
B. Results and Discussion
The result shows that the finetuned VGG16 model can
classify the trash images, including those in the wild, with an
acceptable accuracy of more than 96% despite the
background challenges and the nearby objects that might look
like some other trash classes. However, some
misclassifications still happened, as seen in the confusion
matrix in figure 4.
FIGURE 4: Confusion Matrix for the Classified Trash
Most of the misclassification happened in the trash
class; that occurred because no specific object is
considered in a trash class, as it can contain several
2022 IEEE 12th International Conference on Control System, Computing and Engineering (ICCSCE), 21–22 October 2022, Penang, Malaysia
Authorized licensed use limited to: UNIVERSITI TEKNOLOGI MARA. Downloaded on November 08,2022 at 07:08:42 UTC from IEEE Xplore. Restrictions apply.
different images of different shapes that a model can view
and assign a category to it based on the highest
probability. Similarly, some glass class is misclassified as
plastics or metals; that happened because most of the
glass bottles in the glass class look precisely like plastic
bottles in shape, similar to metal. The same
misclassification occurred with the plastic class with the
glass class. Also, a few misclassifications happened in
other trash classes due to the similarity in shape of some
other objects of the other classes. So, for the model to be
generalized and also minimize misclassification, there is
a need for more pictures, especially in the wild, under
different lighting conditions.
The training and validation accuracy graph is
shown in figure 5; the upper image shows the training
and the validation accuracy steadily increasing based on
the number of epochs. At the same time, the below
image shows the training and validation loss that
declines based on the number of epochs during training
despite the nature of the image samples.
FIGURE 5: Training and Validation Accuracy
In summary, six (6) different trash classes, as in the
trash net dataset, are considered in the attempt to classify
trash images. Half of the images used in this research are from
the trash net dataset, while the remaining images were
obtained from the internet and most are collected manually
from around the university and the neighboring areas. Most
of the images used are trash images in the wild that are very
challenging to be classified by a model. VGG16 model was
finetuned and used for the trash classification, and acceptable
accuracy of more than 96% was achieved within only 20
epochs. The model considers some of the labels on the plastic,
glass, and cardboard trash classes to be papers or trash
This research was funded by Ministry of Higher Education
and USM under LRGS MRUN grant with project code: LRGS
[1] K. N. Sami, Z. M. A. Amin, and R. Hassan, “Waste Management
Using Machine Learning and Deep Learning Algorithms,” Int. J.
Perceptive Cogn. Comput., vol. 6, no. 2, pp. 97–106, Dec. 2020, doi:
[2] S. Shahab, M. Anjum, and M. S. Umar, “Deep Learning Applications
in Solid Waste Management: A Deep Literature Review,” Int. J. Adv.
Comput. Sci. Appl., vol. 13, no. 3, 2022, doi:
[3] M. Triassi, R. Alfano, M. Illario, A. Nardone, O. Caporale, and P.
Montuori, “Environmental Pollution from Illegal Waste Disposal and
Health Effects: A Review on the ‘Triangle of Death,’” Int. J. Environ.
Res. Public. Health, vol. 12, no. 2, pp. 1216–1236, Jan. 2015, doi:
[4] A. A. Namen, F. da Costa Brasil, J. J. G. Abrunhosa, G. G. S.
Abrunhosa, R. M. Tarré, and F. J. G. Marques, “RFID technology for
hazardous waste management and tracking,” Waste Manag. Res. J.
Sustain. Circ. Econ., vol. 32, no. 9_suppl, pp. 59–66, Sep. 2014, doi:
[5] S. Bhandari, “Automatic Waste Sorting In Industrial Environments
Via Machine Learning Approaches,” Master’s Thesis, 2020.
[6] T. Mythili and A. Anbarasi, “A concatenation of deep and texture
features for medicinal trash image classification using EnSegNet-
DNN-based transfer learning,” Mater. Today Proc., vol. 62, pp. 4691–
4698, 2022, doi: 10.1016/j.matpr.2022.03.129.
[7] N. V. Kumsetty, A. Bhat Nekkare, S. K. S., and A. Kumar M.,
“TrashBox: Trash Detection and Classification using Quantum
Transfer Learning,” in 2022 31st Conference of Open Innovations
Association (FRUCT), Helsinki, Finland, Apr. 2022, pp. 125–130. doi:
[8] S. Geetha, J. Saha, I. Dasgupta, R. Bera, I. A. Lawal, and S. Kadry,
“Design of Waste Management System Using Ensemble Neural
Networks,” Designs, vol. 6, no. 2, p. 27, Apr. 2022, doi:
[9] H. Liu, Z. Guo, J. Bao, and L. Xie, “Research on Trash Classification
based on Artificial Intelligence and Sensor,” in 2021 2nd International
Conference on Intelligent Computing and Human-Computer
Interaction (ICHCI), Shenyang, China, Nov. 2021, pp. 274–279. doi:
[10] Y. Yu,A Computer Vision Based Detection System for Trash Bins
Identification during Trash Classification,” J. Phys. Conf. Ser., vol.
1617, no. 1, p. 012015, Aug. 2020, doi: 10.1088/1742-
[11] Datacluster-labs, “Domestic Trash Dataset.” Jul. 2022. Accessed: Aug.
26, 2022. [Online]. Available:
[12] G. Thung and M. Yang, “Classification of Trash for Recyclability
Status,” 2016.
[13] C. Bircanoglu, M. Atay, F. Beser, O. Genc, and M. A. Kizrak,
“RecycleNet: Intelligent Waste Sorting Using Deep Neural
Networks,” in 2018 Innovations in Intelligent Systems and
Applications (INISTA), Thessaloniki, Jul. 2018, pp. 1–7. doi:
[14] Y. Chu, C. Huang, X. Xie, B. Tan, S. Kamal, and X. Xiong,
“Multilayer Hybrid Deep-Learning Method for Waste Classification
and Recycling,” Comput. Intell. Neurosci., vol. 2018, pp. 1–9, Nov.
2018, doi: 10.1155/2018/5060857.
[15] S. K. Zhou et al., A Review of Deep Learning in Medical Imaging:
Imaging Traits, Technology Trends, Case Studies With Progress
Highlights, and Future Promises,” Proc. IEEE, vol. 109, no. 5, pp.
820–838, May 2021, doi: 10.1109/JPROC.2021.3054390.
[16] F. Song, Y. Zhang, and J. Zhang, “Optimization of CNN-based
Garbage Classification Model,” in Proceedings of the 4th
International Conference on Computer Science and Application
Engineering, Sanya China, Oct. 2020, pp. 1–5. doi:
[17] S. Meng and W.-T. Chu, “A Study of Garbage Classification with
Convolutional Neural Networks,” in 2020 Indo Taiwan 2nd
International Conference on Computing, Analytics and Networks
(Indo-Taiwan ICAN), Rajpura, Punjab, India, Feb. 2020, pp. 152–157.
doi: 10.1109/Indo-TaiwanICAN48429.2020.9181311.
[18] J. Bobulski and M. Kubanek, “Deep Learning for Plastic Waste
Classification System,” Appl. Comput. Intell. Soft Comput., vol. 2021,
pp. 1–7, May 2021, doi: 10.1155/2021/6626948.
[19] F. A. Azis, H. Suhaimi, and E. Abas, “Waste Classification using
Convolutional Neural Network,” in Proceedings of the 2020 2nd
International Conference on Information Technology and Computer
2022 IEEE 12th International Conference on Control System, Computing and Engineering (ICCSCE), 21–22 October 2022, Penang, Malaysia
Authorized licensed use limited to: UNIVERSITI TEKNOLOGI MARA. Downloaded on November 08,2022 at 07:08:42 UTC from IEEE Xplore. Restrictions apply.
Communications, Kuala Lumpur Malaysia, Aug. 2020, pp. 9–13. doi:
[20] A. H. Vo, L. Hoang Son, M. T. Vo, and T. Le, “A Novel Framework
for Trash Classification Using Deep Transfer Learning,” IEEE Access,
vol. 7, pp. 178631–178639, 2019, doi:
[21] W. Ma, X. Wang, and J. Yu, “A Lightweight Feature Fusion Single
Shot Multibox Detector for Garbage Detection,” IEEE Access, vol. 8,
pp. 188577–188586, 2020, doi: 10.1109/ACCESS.2020.3031990.
[22] P. Tiyajamorn, P. Lorprasertkul, R. Assabumrungrat, W. Poomarin,
and R. Chancharoen, “Automatic Trash Classification using
Convolutional Neural Network Machine Learning,” in 2019 IEEE
International Conference on Cybernetics and Intelligent Systems (CIS)
and IEEE Conference on Robotics, Automation and Mechatronics
(RAM), Bangkok, Thailand, Nov. 2019, pp. 71–76. doi: 10.1109/CIS-
[23] J. Hong, M. Fulton, and J. Sattar, “A Generative Approach Towards
Improved Robotic Detection of Marine Litter,” in 2020 IEEE
International Conference on Robotics and Automation (ICRA), Paris,
France, May 2020, pp. 10525–10531. doi:
[24] X. Dong, “Research and Design of Marine Trash Classification Robot
Based on Color Recognition,” IOP Conf. Ser. Earth Environ. Sci., vol.
514, no. 3, p. 032043, May 2020, doi: 10.1088/1755-
[25] A. Sun and H. Xiao, “ThanosNet: A Novel Trash Classification
Method Using Metadata,” in 2020 IEEE International Conference on
Big Data (Big Data), Atlanta, GA, USA, Dec. 2020, pp. 1394–1401.
doi: 10.1109/BigData50022.2020.9378287.
[26] F. Alzyoud, W. Maqableh, and F. Al Shrouf, “A Semi Smart Adaptive
Approach for Trash Classification,” Int. J. Comput. Commun.
CONTROL, vol. 16, no. 4, Jul. 2021, doi: 10.15837/ijccc.2021.4.4172.
[27] A. Tatke, M. Patil, A. Khot, P. Jadhav, and D. V. Karad, “HYBRID
Sci. Technol., vol. 5, no. 10, Feb. 2021, doi:
[28] K. Simonyan and A. Zisserman, “Very Deep Convolutional Networks
for Large-Scale Image Recognition.” arXiv, Apr. 10, 2015. Accessed:
Sep. 11, 2022. [Online]. Available:
[29] “garythung/trashnet: Dataset of images of trash; Torch-based CNN for
garbage image classification.”
(accessed Sep. 10, 2022).
2022 IEEE 12th International Conference on Control System, Computing and Engineering (ICCSCE), 21–22 October 2022, Penang, Malaysia
Authorized licensed use limited to: UNIVERSITI TEKNOLOGI MARA. Downloaded on November 08,2022 at 07:08:42 UTC from IEEE Xplore. Restrictions apply.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Conference Paper
The problem of effective disposal of the trash generated by people has rightfully attracted major interest from various sections of society in recent times. Recently, deep learning solutions have been proposed to design automated mechanisms to segregate waste. However, most datasets used for this purpose are not adequate. In this paper, we introduce a new dataset, TrashBox, containing 17,785 images across seven different classes, including medical and e-waste classes which are not included in any other existing dataset. To the best of our knowledge, TrashBox is the most comprehensive dataset in this field. We also experiment with transfer learning based models trained on TrashBox to evaluate its generalizability, and achieved a remarkable accuracy of 98.47%. Furthermore, a novel deep learning framework leveraging quantum transfer learning was also explored. Experimental evaluation on benchmark datasets has shown very promising results. Further, parallelization was incorporated, which helped optimize the time taken to train the models, recording a 10.84% improvement in the performance and 27.4% decline in training time.
Full-text available
Plastic waste management is a challenge for the whole world. Manual sorting of garbage is a difficult and expensive process, which is why scientists create and study automated sorting methods that increase the efficiency of the recycling process. The plastic waste may be automatically chosen on a transmission belt for waste removal by using methods of image processing and artificial intelligence, especially deep learning, to improve the recycling process. Waste segregation techniques and procedures are applied to major groups of materials such as paper, plastic, metal, and glass. Though, the biggest challenge is separating different materials types in a group, for example, sorting different colours of glass or plastics types. The issue of plastic garbage is important due to the possibility of recycling only certain types of plastic (PET can be converted into polyester material). Therefore, we should look for ways to separate this waste. One of the opportunities is the use of deep learning and convolutional neural network. In household waste, the most problematic are plastic components, and the main types are polyethylene, polypropylene, and polystyrene. The main problem considered in this article is creating an automatic plastic waste segregation method, which can separate garbage into four mentioned categories, PS, PP, PE-HD, and PET, and could be applicable on a sorting plant or home by citizens. We proposed a technique that can apply in portable devices for waste recognizing which would be helpful in solving urban waste problems.
Full-text available
Waste management and recycling play a crucial factor in world economy sustainability as they prevent the squander of useful materials which can lead in garbage landfill reduction and cost reduction respectively. Garbage sorting into different categories plays an important role in recycling and waste management; but unfortunately, most garbage sorting still depends on labor which has a reverse impact on mankind and world economy, so there are different approaches to replace human separation by intelligent machines. In this article, we propose a comprehensive approach, Semi Smart Trash Separator to classify garbage and trash using the following technique: precycling by assigning a barcode or QR code to each material, which will enable the separation process as per assigned code; Magnetic separator helps in collecting conductive metal, then the non-conductive materials are classified according to their hardness. This test is a unique idea used in trash classification. Finally, if there is ambiguity in waste material classification barcode or material properties, the classification will be done using neural network techniques depending on the shapes of trash. Mat lab software is modified to handle convolutional neural networks in the image recognition (AlexNet and GoogLeNet) to be used in the trash classification processes and to test their accuracy. The tests are performed using a trustable data set. The material recognition accuracy rate from the obtained results on AlexNet and GoogLeNet are 75% and 83% respectively.
Full-text available
As waste segregation becomes an important issue in our lives, with the use of technology like deep neural networks and computer vision, we can make the process efficient and robust by image segmentation and classification. These systems on the rise need accurate and efficient segmentation and recognition mechanisms and this demand coincides with the increase of computational capabilities of modern computer architectures and more effective algorithms for image recognition. This paper does a comparative analysis of various different approaches and methods like Simple CNN, ResNet50, VGG16, etc in brief. The comparative analysis and study explains the performance of every approach, this paper concludes that ResNet50 gives excellent performance. VGG16 network also provides good performance which meets the needs of daily use.
Solid waste management (SWM) has recently received more attention, especially in developing countries, for smart and sustainable development. SWM system encompasses various interconnected processes which contain numerous complex operations. Recently, deep learning (DL) has attained momentum in providing alternative computational techniques to determine the solution of various SWM problems. Researchers have focused on this domain; therefore, significant research has been published, especially in the last decade. The literature shows that no study evaluates the potential of DL to solve the various SWM problems. The study performs a systematic literature review (SLR) which has complied 40 studies published between 2019 and 2021 in reputed journals and conferences. The selected research studies have implemented the various DL models and analyzed the application of DL in different SWM areas, namely waste identification and segregation and prediction of waste generation. The study has defined the systematic review protocol that comprises various criteria and a quality assessment process to select the research studies for review. The review demonstrates the comprehensive analysis of different DL models and techniques implemented in SWM. It also highlights the application domains and compares the reported performance of selected studies. Based on the reviewed work, it can be concluded that DL exhibits the plausible performance to detect and classify the different types of waste. The study also explains the deep convolutional neural network with the computational requirement and determines the research gaps with future recommendations.
The foremost vital process in clinical trash classification is to classify the medicinal wastes into various categories like contagious, poisonous and normal wastes. For this purpose, several deep learning systems using different structures including ResNext, GoogleNet, etc., are designed. Among those systems, an Enhanced Segmentation Network (EnSegNet) with DNN-TC (EnSegNet-DNN-TC) system has achieved a higher efficiency by segmenting and classifying the trash input images. Although it segments and extracts features effectively, there are very subtle differences between many images because of their highly complex background. This leads to misjudgments of the deep learner system. Therefore in this article, an EnSegNet with Combined Feature Extraction (CFE) and DNN-TC called EnSegNet-CFE-DNN-TC system is proposed to solve misjudgments problem in deep learner classifiers due to high complex background images. First, the input images are segmented by the EnSegNet model. After that, various texture characteristics, namely Gray-Level Co-occurrence Matrix (GLCM), Multi-level Local Binary Pattern (MLBP), Local Derivative Pattern (LDP) and Local Ternary Pattern (LTP) are extracted including the deep features from the segmented images. Then, a new layer called the combination layer is introduced after the Fully Connected (FC) layers to fuse the extracted features and construct a new hybrid feature vector. This hybrid feature vector has a stronger discriminant ability compared to the single feature vector. Further, the softmax is performed to classify the medicinal wastes. Finally, the investigation outcomes reveal that the EnSegNet-CFE-DNN-TC system attains a 93.7% of classification accuracy for 100 trash images compared to the EnSegNet-DNN-TC and DNN-TC.
Since its renaissance, deep learning (DL) has been widely used in various medical imaging tasks and has achieved remarkable success in many medical imaging applications, thereby propelling us into the so-called artificial intelligence (AI) era. It is known that the success of AI is mostly attributed to the availability of big data with annotations for a single task and the advances in high-performance computing. However, medical imaging presents unique challenges that confront DL approaches. In this survey article, we first present traits of medical imaging, highlight both clinical needs and technical challenges in medical imaging, and describe how emerging trends in DL are addressing these issues. We cover the topics of network architecture, sparse and noisy labels, federating learning, interpretability, uncertainty quantification, and so on. Then, we present several case studies that are commonly found in clinical practice, including digital pathology and chest, brain, cardiovascular, and abdominal imaging. Rather than presenting an exhaustive literature survey, we instead describe some prominent research highlights related to these case study applications. We conclude with a discussion and presentation of promising future directions.
Waste Management is one of the essential issues that the world is currently facing does not matter if the country is developed or under developing. The key issue in this waste segregation is that the trash bin at open spots gets flooded well ahead of time before the beginning of the following cleaning process. The isolation of waste is done by unskilled workers which is less effective, time-consuming, and not plausible because of a lot of waste. So, we are proposing an automated waste classification problem utilizing Machine Learning and Deep Learning algorithms. The goal of this task is to gather a dataset and arrange it into six classes consisting of glass, paper, and metal, plastic, cardboard, and waste. The model that we have used are classification models. For our research we did comparisons between four algorithms, those are CNN, SVM, Random Forest, and Decision Tree. As our concern is a classification problem, we have used several machine learning and deep learning algorithm that best fits for classification solutions. For our model, CNN accomplished high characterization on accuracy around 90%, while SVM additionally indicated an excellent transformation to various kinds of waste which were 85%, and Random Forest and Decision Tree have accomplished 55% and 65% respectively