Development of PET Bottle Purchasing Machine using Load Cell and Image Processing with LINE Notify
DOI:
https://doi.org/10.14456/rmutlengj.2021.11Keywords:
Automatic Bottle Purchasing Machine, Weight Sensor (Load Cell), Image processing, MobileNets, LINE Notify, Raspberry PiAbstract
The objective of this research is to develop a prototype machine for purchasing empty plastic bottles of 3 sizes for recycling with notification via LINE Notify. At the initial use, the user must create an account. The user can then sign in and insert empty bottles into the front slot. When a bottle is inserted, the system checks whether the weight is within the specified range by using a weighing sensor. In addition, a camera is used to capture the object’s image and the system determines whether or not it is a bottle using MobileNet-SSD object detector. If not, a servo motor immediately turned over the tray to return that object. Otherwise, the system proceeds to measure the size of the bottle by its width. The weight and size of the bottle are then compared to the pre-defined standard values. A 600 ml bottle should weigh between 60-70 unit (which is read by the HX711 amplifier module) and have a width of 171-184 pixels. A 750 ml bottle should weigh between 85-90 units and have a width of 186-219 pixels and a 1500ml bottle should weigh between 120-150 units and have a width of 226-299 pixels. If the conditions are met, the bottle will be accepted, otherwise it was returned by rotating the tray holding the bottle with a servo motor. After accepting the bottle, the system will update the bottle number, make a payment to the account, as well as send a message to the user through LINE Notify. The experiment results showed that the proposed system was able to accept bottles 26 times out of the 30 tests (an accuracy of about 87 percent). In addition, three different types of erroneous objects were tested for rejection including paper, a PVC pipe and a glue stick for 30 times. The system was able to correctly reject such objects with 100 percent accuracy.
References
[2] Zhao ZQ, Zheng P, Xu ST, Wu X. Object Detection With Deep Learning: A Review. IEEE Transactions on Neural Networks and Learning Systems. 2019;30(11):3212-32. Available from: https://doi.org/10.1109/TNNLS.2018.2876865 [Accessed 11th February 2021].
[3] Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu C-Y, Berg AC. SSD: Single Shot multibox detector. In European conference on Computer vision 2016 Oct 8 (pp.21–37).Springer,Cham.
[4] Lin TY, Maire M, Belongie S, Hays J, Perona P, Ramanan D, Dollar P,Zitnick CL. Microsoft COCO: Common Objects in Context. In European conference on computer vision 2014 Sep 6 (pp.740-755).Springer,Cham.
[5] Howard AG, Zhu M, Chen B, Kalenichenko D, Wang W, Weyand T, Andreetto M, Adam H. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv
preprint arXiv:1704.04861. 2017 Apr 17.
[6] Padilla R, Netto SL, da Silva EA. A Survey on performance metrics for object-detection algorithms. In 2020 International Conference on Systems, Signals and Image Processing (IWSSIP)
2020 Jul 1 (pp. 237–242). IEEE.
[7] Chattoraj S, Bhowmik S, Vishwakarma K, Roy P. Design and implementation of low cost electronic toll collection system in India. 2017 Second International Conference on Electrical,
Computer and Communication Technologies (ICECCT)2017 Feb 22 (pp. 1–4). IEEE.
For Author
