ABSTRACT
At the moment, 2.01 billion tons of solid trash are produced annually in all cities throughout the world. Based on research findings, a typical person generates approximately 0.74 kilograms of garbage on a daily basis. The researchers decided to help the community by providing information on household waste detection and classification. By means of implementing the Mask R-CNN framework, it became feasible to obtain a segmented mask as well as a bounding box for every identified household waste item in a photo. Bounding boxes were used to identify multiple wastes, things with different scales, and objects that overlap in an image. Bounding boxes were made of various household wastes. The wastes were classified into biodegradable, non-biodegradable, and recyclable which has the potential to mitigate human labor and expedite the waste segregation process. The Mask R-CNN model was used to improve detection and demonstrate how household trash may be appropriately identified in a photo with several objects. The model for detecting objects encountered challenges in detecting household wastes caused by their deformity, cluttered and occluded environment, and poor lighting conditions. The Mask R-CNN model produced a 38.25% mean average precision, 65.50% mean average recall, and 48.29% F1 score. The proposed model can be utilized in detecting and classifying household wastes.
- Anon. What a waste 2.0. Retrieved February 12, 2022 from https://datatopics.worldbank.org/what-a-waste/trends_in_solid_waste_management.htmlGoogle Scholar
- Jonathan L. Mayuga. 2019. Metro Manila producing more garbage every year–DENR. (December 2019). Retrieved February 22, 2022 from https://businessmirror.com.ph/2019/12/24/metro-manila-producing-more-garbage-every-year-denr/Google Scholar
- Olugboja Adedeji and Zenghui Wang. 2019. Intelligent Waste Classification system using deep learning convolutional neural network. Procedia Manufacturing 35 (2019), 607–612. DOI:https://doi.org/10.1016/j.promfg.2019.05.086Google ScholarCross Ref
- Yujie He, Qinyue Gu, and Maguo Shi. 2020. Trash Classification Using Convolutional Neural Networks Project Category: Computer Vision. (2020). Retrieved February 12, 2022 from http://cs230.stanford.edu/projects_spring_2020/reports/38847029.pdfGoogle Scholar
- Yushuo Chang. 2022. Waste Detection Based On Mask R-CNN. (July 2022). Retrieved March 12, 2023 from https://webthesis.biblio.polito.it/23504/Google Scholar
- Yang Yu, Kailiang Zhang, Li Yang, and Dongxing Zhang. 2019. Fruit detection for strawberry harvesting robot in non-structural environment based on mask-RCNN. Computers and Electronics in Agriculture 163 (2019), 104846. DOI:http://dx.doi.org/10.1016/j.compag.2019.06.001Google ScholarDigital Library
- Farzana Shaikh, Nagma Kazi, Farheen Khan, and Zaid Thakur. 2020. Waste profiling and analysis using machine learning. 2020 Second International Conference on Inventive Research in Computing Applications (ICIRCA) (2020). DOI:http://dx.doi.org/10.1109/icirca48905.2020.9183035Google ScholarCross Ref
- Punitha Jaikumar, Remy Vandaele, and Varun Ojha. 2021. Transfer learning for instance segmentation of waste bottles using mask R-CNN algorithm. Advances in Intelligent Systems and Computing (2021), 140–149. DOI:http://dx.doi.org/10.1007/978-3-030-71187-0_13Google ScholarCross Ref
- Kaiming He, Georgia Gkioxari, Piotr Dollar, and Ross Girshick. 2017. Mask R-CNN. 2017 IEEE International Conference on Computer Vision (ICCV) (2017). DOI:http://dx.doi.org/10.1109/iccv.2017.322Google ScholarCross Ref
- Ahsen Battal and Adem Tuncer. 2022. Detection of Face Mask Wearing Condition for COVID-19 using Mask R-CNN. El-Cezerî Journal of Science and Engineering (2022). DOI:http://dx.doi.org/10.31202/ecjse.1061270Google ScholarCross Ref
- Hend Bakr, Ahmed Hamad, and Khalid Amin. 2020. Mask R-CNN for moving shadow detection and segmentation. IJCI. International Journal of Computers and Information 0, 0 (2020), 0–0. DOI:http://dx.doi.org/10.21608/ijci.2020.44215.1029.Google ScholarCross Ref
- Rufus Rubin, Chinnu Jacob, S.M. Anzar, and Alavikunhu Panthakkan. 2022. Mask R-CNN with Multi-Backbones - A Comparative Analysis. 2022 5th International Conference on Signal Processing and Information Security (ICSPIS) (2022). DOI:http://dx.doi.org/10.1109/icspis57063.2022.10002546Google ScholarCross Ref
- Arun Kumar Singh, Arun Kant Dwivedi, Maram Sumanth, and Dharmendra Singh. 2022. An Efficient Approach for Instance Segmentation of Railway Track Sleepers in Low Altitude UAV Images Using Mask R-CNN. IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium (2022). DOI:http://dx.doi.org/10.1109/igarss46834.2022.9883474Google ScholarCross Ref
- Ahyun Lee and Insung Jang. 2018. Mouse picking with Ray casting for 3D spatial information open-platform. 2018 International Conference on Information and Communication Technology Convergence (ICTC) (2018). DOI:http://dx.doi.org/10.1109/ictc.2018.8539567Google ScholarCross Ref
- Usag1r. 2022. How to batch resize multiple images in Python (via Pil Library). (August 2022). Retrieved March 12, 2023 from https://holypython.com/how-to-batch-resize-multiple-images-in-python-via-pil-library/Google Scholar
- Bernardo Caldas. 2020. Using MASKRCNN to predict tropical fruits in custom dataset. (January 2020). Retrieved December 10, 2022 from https://medium.com/analytics-vidhya/using-maskrcnn-to-predict-tropical-fruits-in-custom-dataset-4f079d05fbe1Google Scholar
- Dianne Castillo. 2021. Transfer Learning for Machine Learning. (June 2021). Retrieved February 22, 2022 from https://www.seldon.io/transfer-learning#:∼:text=Transfer%20learning%20means%20taking%20the,to%20solve%20a%20specific%20taskGoogle Scholar
- Anon. Confusion matrix. Retrieved December 10, 2022 from https://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html#:∼:text=The%20diagonal%20elements%20represent%20the,better%2C%20indicating%20many%20correct%20predictionsGoogle Scholar
Index Terms
- Multiple Garbage Detection using Mask R-CNN in TensorFlow for Household Waste Classification
Recommendations
Waste Classification from Digital Images Using ConvNeXt
Image and Video TechnologyAbstractIn this paper, ConvNeXt is selected as a model for waste classification from digital images. ConvNeXt is a CNN-based backbone network that has been proposed to further improve the performance of models for visual tasks, following the various types ...
Application of generated mask method based on Mask R-CNN in classification and detection of melanoma
Highlights- A melanoma detection framework Mask-DenseNet+ is proposed based on the Mask-RCNN. It combines Mask R-CNN networks to add mask processing operations, ...
Abstract ObjectiveMelanoma is a type of malignant skin cancer with high mortality, and its incidence is increasing rapidly in recent years. At present, the best treatment is surgical resection after early diagnosis. However, due to ...
Radio Galaxy Morphology Classification with Mask R-CNN
ICVISP 2020: Proceedings of the 2020 4th International Conference on Vision, Image and Signal ProcessingDue to the sheer scale of data generated from newer radio galaxy surveys, automating the classification of radio galaxy morphologies has become an active area of study in recent years. One promising solution is CLARAN: A proof-of-concept radio source ...
Comments