No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Domestic Trash Classification with Transfer Learning Using VGG16
... The proposed work is discussed, its results are compared with current approaches, and the relevance of the CNN-GRU model that utilizes transfer learning in the classification of waste is examined. The results of the study showed that the combination CNN-GRU model with transfer learning outperformed the traditional CNN and LSTM models, which obtained accuracies of 89% and 92%, respectively, with a superior accuracy of 97% [17]. Regularization was an important factor that helped in enhancing the classification outcomes through the assistance of transfer learning that helped the model in predecessor knowledge from the pretrained networks on big data sets that acts as a strong baseline or a prior distribution used for feature extraction. ...
Waste classification remains pivotal to environmental sustainability along with proper waste management. Traditional approaches such as CNNs and LSTM networks prove to be inadequate in properly capturing the spatial and temporal correlations in waste images. To cover for this, the study puts forward a new waste classification strategy that leverages CNNs, GRUs and transfer learning for increased classification performance. It is worth mentioning that proposed approach relies on CNNs for the spatial feature extraction, GRUs for temporal sequence learning and transfer learning for utilizing pre-trained models for both feature extraction and sequential learning. The proposed framework is developed in Python and tested on the waste classification dataset with accuracy of 97% which is superior to the traditional CNN (89%) and LSTM (92%). This result underlines the capability of applying the transfer learning of convolution neural network (CNN-GRU) that has the capability to develop an effective framework for waste classification efficiently. The research finds that use of such techniques enhances development of AI based solutions towards efficient waste management and environment protection.
... Yasaswini et al. [25] identified the model to avoid road accidents using DL models. Abdu and Noor [26] presented a VGG16-based DL model for domestic trash classification to classify trash items, achieving over 96% precision appropriately. Shyam et al. [27] suggested an unsupervised domain adaptation technique that leverages a simulator-created synthetic dataset to ensure performance consistency of multi-object-tracking (MOT) algorithms over various manually annotated real situations. ...
As humans, we can easily recognize and distinguish different features of objects in images due to our brain’s ability to unconsciously learn from a set of images. The objectives of this effort are to develop a model that is capable of identifying and categorizing objects that are present within images. We imported the dataset from Keras and loaded it using data loaders to achieve this. We then utilized various deep learning algorithms, such as visual geometry group (VGG)-16 and a simple net-random forest hybrid model, to classify the objects. After classification, the accuracy obtained by VGG16 and the hybrid model was 84.7% and 89.6%, respectively. Therefore, the proposed model successfully detects objects in images using a simple net as a feature extractor and a random forest for object classification, achieving better accuracy than VGG16.
... With the rising urbanization of the India presents countless dangers likewise with expansion in populace land utilization increments, utility increments, utilization of food expands, utilization of assets increments, and more than these the amount of waste created by 1.37 billion individuals increments [1]. Squander the board framework is difficult for metropolitan regions among most pieces of nations all around the world [2]. ...
Authorities in non-industrial nations like India ordinarily recognize the requirement for better executives. Nonetheless, little endeavors are finished to advance the circumstance, changes usually take long-time frame. We probably are aware, India as population that is comparable to 17.7% of the complete population. As smart cite development is increasing in India, a classification system is much needed. Since how much waste is increasing step by step. The present items in India rehearse an assortment of the homegrown, modern litter are unloaded into large free areas. The litter detachment is carried out by the workers that are not efficient, consume a great deal of time, and aren't even totally doable because of a lot of litter. The motivation behind this examination is to construct a continuous application that perceives the kind of squander and classifies it into characterized classifications. The particular framework guarantees the most effective way to squander the executives and will likewise accelerate the isolation interaction with higher exactness. This study endures surprising results and is fruitful to group different pictures of waste in the right classes.
... With the rising urbanization of the India presents countless dangers likewise with expansion in populace land utilization increments, utility increments, utilization of food expands, utilization of assets increments, and more than these the amount of waste created by 1.37 billion individuals increments [1]. Squander the board framework is difficult for metropolitan regions among most pieces of nations all around the world [2]. ...
... Since it is costly and time-consuming to separate trash manually, it is very important to develop an automatic system for separate collection using deep learning and computer vision [1][2][3][4]. In previous research on automatic systems for separate collection, many methods have been proposed for trash classification using deep learning and computer vision [5][6][7]. However, they only deal with the simple case for the image with a single trash. ...
Due to the sharp increase in household waste, its separate collection is essential in order to reduce the huge amount of household waste, since it is difficult to recycle trash without separate collection. However, since it is costly and time-consuming to separate trash manually, it is crucial to develop an automatic system for separate collection using deep learning and computer vision. In this paper, we propose two Anchor-free-based Recyclable Trash Detection Networks (ARTD-Net) which can recognize overlapped multiple wastes of different types efficiently by using edgeless modules: ARTD-Net1 and ARTD-Net2. The former is an anchor-free based one-stage deep learning model which consists of three modules: centralized feature extraction, multiscale feature extraction and prediction. The centralized feature extraction module in backbone architecture focuses on extracting features around the center of the input image to improve detection accuracy. The multiscale feature extraction module provides feature maps of different scales through bottom-up and top-down pathways. The prediction module improves classification accuracy of multiple objects based on edge weights adjustments for each instance. The latter is an anchor-free based multi-stage deep learning model which can efficiently finds each of waste regions by additionally exploiting region proposal network and RoIAlign. It sequentially performs classification and regression to improve accuracy. Therefore, ARTD-Net2 is more accurate than ARTD-Net1, while ARTD-Net1 is faster than ARTD-Net2. We shall show that our proposed ARTD-Net1 and ARTD-Net2 methods achieve competitive performance in mean average precision and F1 score compared to other deep learning models. The existing datasets have several problems that do not deal with the important class of wastes produced commonly in the real world, and they also do not consider the complex arrangement of multiple wastes with different types. Moreover, most of the existing datasets have an insufficient number of images with low resolution. We shall present a new recyclables dataset which is composed of a large number of high-resolution waste images with additional essential classes. We shall show that waste detection performance is improved by providing various images with the complex arrangement of overlapped multiple wastes with different types.
Municipal solid waste (MSW) accumulation is a critical global challenge for society and governments, impacting environmental and social sustainability. Efficient separation of MSW is essential for resource recovery and advancing sustainable urban management practices. However, manual classification remains a slow and inefficient practice. In response, advances in artificial intelligence, particularly in machine learning, offer more precise and efficient alternative solutions to optimize this process. This research presents the development of a light deep neural network called R3sNet (three “Rs” for Reduce, Reuse, and Recycle) with residual modules trained end-to-end for the binary classification of MSW, with the capability for faster inference. The results indicate that the combination of processing techniques, optimized architecture, and training strategies contributes to an accuracy of 87% for organic waste and 94% for inorganic waste. R3sNet outperforms the pre-trained ResNet50 model by up to 6% in the classification of both organic and inorganic MSW, while also reducing the number of hyperparameters by 98.60% and GFLOPS by 65.17% compared to ResNet50. R3sNet contributes to sustainability by improving the waste separation processes, facilitating higher recycling rates, reducing landfill dependency, and promoting a circular economy. The model’s optimized computational requirements also translate into lower energy consumption during inference, making it well-suited for deployment in resource-constrained devices in smart urban environments. These advancements support the following Sustainable Development Goals (SDGs): SDG 11: Sustainable Cities and Communities, SDG 12: Responsible Consumption and Production, and SDG 13: Climate Action.
Waste classification remains pivotal to environmental sustainability along with proper waste management. Traditional approaches such as CNNs and LSTM networks prove to be inadequate in properly capturing the spatial and temporal correlations in waste images. To cover for this, the study puts forward a new waste classification strategy that leverages CNNs, GRUs and transfer learning for increased classification performance. It is worth mentioning that proposed approach relies on CNNs for the spatial feature extraction, GRUs for temporal sequence learning and transfer learning for utilizing pre-trained models for both feature extraction and sequential learning. The proposed framework is developed in Python and tested on the waste classification dataset with accuracy of 97% which is superior to the traditional CNN (89%) and LSTM (92%). This result underlines the capability of applying the transfer learning of convolution neural network (CNN-GRU) that has the capability to develop an effective framework for waste classification efficiently. The research finds that use of such techniques enhances development of AI based solutions towards efficient waste management and environment protection.
Waste management is an essential societal issue, and the classical and manual waste auditing methods are hazardous and time-consuming. In this paper, we introduce a novel method for waste detection and classification to address the challenges of waste management. The method uses a collection of deep neural networks to allow for accurate waste detection, classification, and waste size quantification. The trained neural network model is integrated into a mobile-based application for trash geotagging based on images captured by users on their smartphones. The tagged images are then connected to the cleaners’ database, and the nearest cleaners are notified of the waste. The experimental results using publicly available datasets show the effectiveness of the proposed method in terms of detection and classification accuracy. The proposed method achieved an accuracy of at least 90%, which surpasses that reported by other state-of-the-art methods on the same datasets.
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.
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.
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.
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.
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.
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
To solve the problems of a poor manual garbage sorting environment, including heavy tasks and low sorting efficiency, we propose the Lightweight Feature Fusion Single Shot Multibox Detector (L-SSD) algorithm to realize intelligent trash classification and recognition. Since waste has a small volume and the image resolution of garbage is always low, the algorithm that we propose is an enhanced single shot multibox detector (SSD) with a lightweight and novel feature fusion module. This SSD can significantly improve the performance of rubbish detection. In this feature fusion module, features from different layers with different scales are connected in series. A new feature pyramid was generated by using downsampling blocks, which will be fed to appointed multibox detectors to predict the final detection results. Due to the extremely unbalanced ratio of positive samples to negative samples, which leads to a low accuracy of SSD, Focal Loss using balanced cross-entropy is employed, which is provided by easy examples that corresponds to difficult samples with a decline in the loss weight. Thus, the training is biased towards meaningful samples. We have replaced the backbone network of VGG16 with ResNet-101 to achieve more accurate detection. We analyzed the performance of a nonmaximum suppression (NMS) algorithm and discovered that Soft-NMS was more suitable for learning better image representations. The strategy of Soft-NMS is to suppress the undesirable detection box rather than remove it completely. The experimental results show that the L-SSD exceeds a large number of state-of-the-art object detection algorithms in both accuracy and speed.
Nowadays, society is growing and crowded, the construction of automatic smart waste sorter machine utilizing the intelligent sensors is important and necessary. To build this system, trash classification from trash images is an important issue in computer vision to be addressed for integrating into sensors. Therefore, this study proposes a robust model using deep neural networks to classify trash automatically which can be applied in smart waste sorter machines. Firstly, we collect the VN-trash dataset that consists of 5904 images belonging to three different classes including Organic, Inorganic and Medical wastes from Vietnam. Next, this study develops a deep neural network model for trash classification named DNN-TC which is an improvement of ResNext model to improve the predictive performance. Finally, the experiments are conducted to compare the performances of DNN-TC and the state-of-the-art methods for trash classification on VN-trash dataset as well as Trashnet dataset to show the effectiveness of the proposed model. The experimental results indicate that DNN-TC yields 94% and 98% in terms of accuracy for Trashnet and VN-trash datasets respectively and thus it outperforms the state-of-the-art methods for trash classification on both experimental datasets.
In this study, an image based detection system is designed and applied to identify different trash bins for classification. The CCD industrial camera is used to obtain the RGB format images on the trash disposal line in real time. In order to reduce the effect of environmental brightness, the obtained RGB format images are transformed into HSI color format. According to the calculated saturation and hue values, the threshold of saturation and hue values is set for the following image segmentation, respectively. Then, accurate position information of trash bin is obtained for the subsequent classification processing. This detection system could overcome environmental constraints and improve efficiency for trash classification.
The UN (United Nations) identifies sustainable development of the marine ecosystem as a major goal. To achieve the goal, the garbage in the ocean has to be removed, and as people recognize that different types of classification should be disposed differently, classification is critical after garbage is collected. This led to the engineering goal of constructing a robot that can pick up garbage in the water and classify the garbage into its right category. The robot was developed by attaining four critical functions: driving under water condition, a set of gripper, color recognition and garbage classification. Results of data obtained from developing the garbage classification was plotted on multiple graphs including training and validation loss, training time, training and testing accuracy and so on. After the critical functions are fulfilled, it is shown that the accumulative testing accuracy for garbage classification algorithm was around 90.6%, while the programs for the other three critical functions all compiled successfully. It was a regret that datasets weren’t shaped to the same sizes and the critical functions should be synthesized for further research.
This study proposes a multilayer hybrid deep-learning system (MHS) to automatically sort waste disposed of by individuals in the urban public area. This system deploys a high-resolution camera to capture waste image and sensors to detect other useful feature information. The MHS uses a CNN-based algorithm to extract image features and a multilayer perceptrons (MLP) method to consolidate image features and other feature information to classify wastes as recyclable or the others. The MHS is trained and validated against the manually labelled items, achieving overall classification accuracy higher than 90% under two different testing scenarios, which significantly outperforms a reference CNN-based method relying on image-only inputs.
Waste management and recycling is the fundamental part of a sustainable economy. For more efficient and safe recycling, it is necessary to use intelligent systems instead of employing humans as workers in the dump-yards. This is one of the early works demonstrating the efficiency of latest intelligent approaches.
In order to provide the most efficient approach, we experimented on well-known deep convolutional neural network architectures. For training without any pretrained weights, Inception-Resnet, Inception-v4 outperformed all others with 90\% test accuracy. For transfer learning and fine-tuning of weight parameters using ImageNet, DenseNet121 gave the best result with 95\% test accuracy. One disadvantage of these networks, however, is that they are slightly slower in prediction time. To enhance the prediction performance of the models we altered the connection patterns of the skip connections inside dense blocks.
Our model RecycleNet is carefully optimized deep convolutional neural network architecture for classification of selected recyclable object classes. This novel model reduced the number of parameters in a 121 layered network from 7 million to about 3 million.
The term "triangle of death" was used for the first time by Senior and Mazza in the journal The Lancet Oncology referring to the eastern area of the Campania Region (Southern Italy) which has one of the worst records of illegal waste dumping practices. In the past decades, many studies have focused on the potential of illegal waste disposal to cause adverse effects on human health in this area. The great heterogeneity in the findings, and the bias in media communication has generated great healthcare doubts, anxieties and alarm. This paper addresses a review of the up-to-date literature on the "triangle of death", bringing together the available information on the occurrence and severity of health effects related to illegal waste disposal. The Scopus database was searched using the search terms "waste", "Campania", "Naples", "triangle of death" and "human biomonitoring". Despite the methodological and sampling heterogeneity between the studies, this review examines the evidence from published data concerning cancer incidence, childhood mortality and birth defects, so that the current situation, knowledge gaps and research priorities can be established. The review aims to provide a contribution to the scientific community, and to respond to the concerns of the general population.
The illegal dumping of hazardous waste is one of the most concerning occurrences related to illegal waste activities. The waste management process is quite vulnerable, especially when it comes to assuring the right destination for the delivery of the hazardous waste. The purpose of this paper is to present a new system design and prototype for applying the RFID technology so as to guarantee the correct destination for the hazardous waste delivery. The aim of this innovative approach, compared with other studies that employ the same technology to the waste disposal process, is to focus on the certification that the hazardous waste will be delivered to the right destination site and that no inappropriate disposal will occur in the transportation stage. These studies were carried out based on data collected during visits to two hazardous waste producer companies in Brazil, where the material transportation and delivery to a company in charge of the waste disposal were closely monitored.
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.
In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively.
In this work we investigate the effect of the convolutional network depth on
its accuracy in the large-scale image recognition setting. Our main
contribution is a thorough evaluation of networks of increasing depth, which
shows that a significant improvement on the prior-art configurations can be
achieved by pushing the depth to 16-19 weight layers. These findings were the
basis of our ImageNet Challenge 2014 submission, where our team secured the
first and the second places in the localisation and classification tracks
respectively.
Automatic Waste Sorting In Industrial Environments Via Machine Learning Approaches
- S Bhandari
S. Bhandari, "Automatic Waste Sorting In Industrial Environments
Via Machine Learning Approaches," Master's Thesis, 2020.
Domestic Trash Dataset
- Datacluster-Labs
Datacluster-labs, "Domestic Trash Dataset." Jul. 2022. Accessed: Aug.
26, 2022. [Online]. Available: https://github.com/dataclusterlabs/Domestic-Trash-Dataset
Classification of Trash for Recyclability Status
- G Thung
- M Yang
G. Thung and M. Yang, "Classification of Trash for Recyclability
Status," 2016.