Automatic Trash Recognition System Using Artificial Intelligence and Machine Learning
for Avoid Covid -19
Wiki Kusno1*, Indra Maulana1, Zulfykar1, Ujang Sudrajat1, Ari Purno Wahyu W1 and Zairi Ismael Rizman2
1Informatics Department, Faculty of Engineering, Widyatama University, Bandung, Indonesia
2School of Electrical Engineering, College of Engineering, Universiti Teknologi MARA, 23000 Dungun,
Terengganu, Malaysia
*wiki.kusno@widyatama.ac.id
Article History: Received: 10 January 2021; Revised: 12 February 2021; Accepted: 27 March 2021; Published
online: 28 April 2021
Abstract: The covid-19 period that emerged in March 2020 is still a big problem in various countries to date, the
impact is quite large and has resulted in several sectors of the country's economy paralyzed, currently the government is trying to prevent it in various ways, starting with healthy socialization to implementing scale lockdown with limited namely between regions and cities, the spread of Covid which starts only through the air turns into spread through objects around us, so that spraying is often carried out by infecting or using a hand sanitizer, the two methods are currently unable to fully address prevention Covid but by reducing the spread so that it doesn’t spread again, the most important thing is how we can live healthy and clean, in this study one of the prevention steps is to use smart waste technology which is expected enable prevent the spread of Covid -19 and declare Specify the waste into two categories of metal and non-metal, this is condition because Covid itself can spread through the air, especially through garbage, this technology uses the help of electronic sensors and computerized technology where the type of waste will be classified and then sorted with a conveyor machine so that non-waste organic matter doesn’t experience decomposition and spreads viruses, the accuracy of the reading from this sensor depends on the amount of training data used, while the processing of waste images uses Neural-Network and artistic intelligence algorithms with an accuracy of reading data up to 95%.
Keywords: Waste sorting, artificial intelligence, neural networks, image processing
I. INTRODUCTION
The spread of viruses and hazardous substances can be transmitted by waste that isn’t properly wasted, especially if the garbage is in a densely populated area so that the process of spreading viruses such as covid-19 and other viruses can spread quickly through the air and objects around us either metal or non-metal objects, the virus non-metal objects can last longer and are dangerous if touched with bare hands, a waste disposal management is needed so that waste doesn’t accumulate and is disposed of regularly.
The COVID-19 crisis affects not only greenhouse gases and energy, but also waste. The more time we spend at home; the more rubbish we remove from our homes. Increasing the number of home deliveries increases the number of boxes and other emissions. It also reduces business waste from shops and offices. Careful handling of infectious waste is also required. Another surge in demand for masks and disposable test kits has also increased the amount of contagious plastic waste thrown away. In Japan, the maintenance of waste treatment and recycling infrastructure has become a major problem as the population declines and the embargo on plastic waste is in place in neighboring countries. The COVID-19 crisis It can be said that this adds to a significant problem situation [1].
In the case of COVID-19 as reported by the World Health Organization (WHO), “Situation Report-12”, the mode of transmission could be similar to the previous epidemic caused by other people Coronavirus (MERS, Middle East Respiratory Syndrome and SARS, Acute Respiratory Syndrome), where person-to-person transmission occurs via droplets, aerosols and direct contact [2].
II. LITERATURE REVIEW
The process of disposing of waste in dense areas has become the concern of several parties at this time, where a clean and tidy environment will affect our immune system and health, waste that is allowed to accumulate and isn’t sorted into toxic and dangerous, the process of sorting waste into a part important which usually uses manual by hand then sorted using electronic and mechanical systems, computer techniques have also been developed for waste sorting systems and combining the concept of a neural-network algorithm, this algorithm has a high data reading accuracy that can clarify the types of waste glass, metal, paper and plastic with an accuracy of reading data up to 87% [3].
Based on a study conducted by the World Bank, a densely populated settlement contributes more than 4 million tons of waste worldwide and is predicted to increase in 2025 and will increase in developing countries in the next 25 years, this is due to the side effects of industrial development that causes increased waste pollution, if left unchecked it will cause pollution and damage to water resources and the environment and if spread in the open air can cause cancer, some researchers have tried to develop how to recycle waste in a safe and economical way [4].
In a conventional method, the waste selection process is carried out to group and classify an object, while the problem that arises from this method is that people who carry out the sorting process will be exposed to hazardous waste and substances, so we need an automatic method that is able to work quickly, large scale sorting system of waste can be recycled and used as an energy source, the waste sorting system was further developed using an SVM method combined with a neural-network where this accuracy and method is able to recognize and classify types of waste [5].
It should be noted that most of the waste generated in large cities can be recycled. Identification and implementation of methods that can benefit or at least mitigate the environment. There are techniques or models that help people sort waste to be important in disposing of these materials properly. Even though there are different types of recycling categories, people may still be confused or not well recognized as to how to find the right container to dispose of any trash.
To further minimize the effects of improper disposal of domestic waste (eg paper, plastics, glass, and waste), an automated system based on neural network techniques that aims for the correct separation of waste in the recycling category. The way people handled trash in the past for centuries it was still based on original strategies for and improving hygiene. Population growth is an important factor in increasing the production of this waste. That is why it must be reduced to maintain a balance in waste management [6]. Garbage has become a major problem worldwide due to the uncontrolled disposal of household, household and industrial residents without waste that is effective and efficient, management programs that involve health risks and can have a negative impact on the environment Efficient classification of waste management plays a role important in ecology, sustainable development by ensuring proper, efficient waste treatment processes Selective collection is often used to improve recycling and protect the environment in developing countries where waste management is a serious problem for economic development [7].
The current waste recycling process requires several recycling plants to handle waste and uses a combination of filters and large tools to separate objects of a certain shape. Improving this recycling process will help increase plant efficiency by reducing waste and reducing the time it takes to sort waste. this process increases the accuracy of the classification rather than is advantageous from both an environmental and economic point of view. A waste image is fed to a system similar to that used in a waste disposal facility so that the waste data is sorted. Different objects in the image are identified and
each object is classified separately. Image processing algorithms are used to classify images into six different categories. By automating this process, we can easily classify objects that can be recycled just by looking at the image [8].
III. METHODOLOGY
In this experiment, the authors used several important stages, namely data visualization, image processing and finally identification using algorithms, these stages are as follows.
Figure 1.1. Reading of garbage objects by the system
In Figure 1.1 is a process of reading trash objects that are taken directly by the camera using the C programming language, in that mode the camera object used can be different - with different resolutions so that the system will make it a standard size of 400 x 200 pixels, this level of resolution will affect the process of reading the data, when taking pictures of trash objects will vary because it is seen in terms of color and size, the system will then convert it to black and white.
Figure 1.2. Object reading with bounding box
In Figure 1.2 is a process of reading the image by providing a marking mark or "bounding box" which serves to mark the detected garbage object.
Figure 1.3. Reading an object with a bounding box
In Figure 1.3 is a process of reading data by changing the image color to greyscale by using two variable
values "zeros" and "area".
In the picture above, an experiment is detected using a garbage data folder that is already available, this experiment is carried out to check the accuracy of the data reading carried out by the system, a program is said to be successful if the accuracy is above 50% and the error value is small. The data used in this study are types of organic and non-organic waste, with the number of datasets for each sample of up to 100 images and again classified as plastic, glass and paper waste.
IV. RESULTS AND DISCUSSION
In this experiment, two types of data were used, namely training data and testing data, where both data were used to increase the accuracy value of the system reading process, the waste samples used were plastic waste and metal waste.
Figure 1.4. Sample data of bottle waste
In Figure 1.4 shows several samples of waste, namely metal and non-metal, while in this experiment the detection process used non-metal or glass waste, data collection using a camera input with the shooting position at the top, this retrieval process is carried out so that the object can be seen clearly and reduce the reading error which will cause the accuracy of the reading to decrease.
In Figure 1.5 is a visual classification process where a detected garbage data is read by an algorithm to separate the original object and the background so that the level of accuracy is higher, this algorithm reads objects based on shape, color and size.
Figure 1.6. Greyscale process
In Figure 1.6 is a color visualization method, this method is useful for seeing a waste object that is categorized as metal or non-metal and is useful in a random sorting system, in the experiment above, it can be seen that the bottle waste has a different color from the surrounding background.
Figure 1.7. Results of object detection
In Figure 1.7 is the final result of reading the system, visually the algorithm will provide a color mark to distinguish the type of waste and this method is useful in the sorting process randomly and in large quantities.
Figure 1.8. Detection of garbage cans
In Figure 1.8 is an experiment that is used on non-metal waste with the results of reading the accuracy of the system reaching 95% and is included in the category of non-organic waste or into the classification of metal waste.
Figure 1.9. A mixture of metal and non-metal waste
In Figure 1.9 is an experiment on mixed waste, this mixed experiment is used to prove whether or not the application can read types of metal and non-metal waste from the test results proven to be metal waste with a reading accuracy of up to 98%.
Figure 1.10. Detection of non-organic waste
In Figure 1.10 is an experiment with computer technology that is used to detect types of metal and non-metallic waste, the ability to read data accuracy depends on the number of datasets used, to increase the accuracy value of waste data used to reach more than 100 datasets of plastic waste and non-plastic, the visual results show that the computer can distinguish types of waste and display in detail with an accuracy of above 90%.
V. CONCLUSION
From the results of the research, it can be concluded that a waste sorting system is very necessary, especially if the waste is in large quantities which, if left untreated, can decompose and cause soil and water pollution, the previous solution was the sorting system using human assistance with the help of hands and other personal protective equipment, but this method has several risks, namely being able to be exposed to sorting tasks especially if the waste contains hazardous chemicals, in developed countries waste sorting is done by sorting organic and non-organic waste where non-organic waste itself can be recycled and has a high economic value, sorting system which in this experiment is to use image processing technology based on artificial intelligence where a system is able to distinguish types of waste objects and is able to read an accuracy rate of reading data up to 90%, for further development this system can be used conventional machines, which has a fast and large number of sorting systems, the process of increasing the accuracy improve able increasing the number of datasets so that data on various items and types of waste can be read by the system, this sorting system can be used to prevent the Covid-19 virus and other viruses from spreading. so that officers and the general public are not exposed to hazardous substances.
References
[1] Onoda, H. (2020). Smart approaches to waste management for post-COVID-19 smart cities in Japan. IET Smart Cities, 2(2), 89-94.
[2] World Health Organization (WHO). "WHO Novel Coronavirus (2019-nCoV)," WHO, Geneva, 2020. [3] Adedeji, O., & Wang, Z. (2019). Intelligent waste classification system using deep learning
convolutional neural network. Procedia Manufacturing, 35, 607-612.
[4] Hoornweg, D., & Bhada-Tata, P. (2016). What a waste: A global review of solid waste management. Washington DC: World Bank Group, 2012. pp. 98. Urban Development Series Knowledge Papers. [5] Johansson, N., & Corvellec, H. (2018). Waste policies gone soft: An analysis of European and
Swedish waste prevention plans. Waste Management, 77, 322-332.
[6] Islam, M. S., Arebey, M., Hannan, M. A., & Basri, H. (2012). Overview for solid waste bin monitoring and collection system. IEEE International Conference on Innovation Management and Technology Research, pp. 258-262.
[7] Arebey, M., Hannan, M. A., Basri, H., Begum, R. A., & Abdullah, H. (2011). Integrated technologies for solid waste bin monitoring system. Environmental Monitoring and Assessment, 177(1), 399-408. [8] Prasanna, A., & Vikash Kaushal, S. (2018). Survey on identification and classification of waste for
