Garbage piles up faster than any Indian municipal-ity can deal with it. The usual approach to understanding what people throw away still involves workers tearing open bags by hand, which is both slow and genuinely hazardous. Most attempts to automate the process have relied on convolutional neural networks that slap a single label onto an entire photograph. That is adequate for a picture of one bottle on a table. It is useless for a picture of actual roadside garbage where a squashed cola can sits on top of yesterday’s newspaper next to a torn polythene bag. What you really want is a system that can look at the heap and individually outline each piece of rubbish, separately. This paper describes EcoMask, a system we built around Mask R-CNN to do exactly that. We trained on the TACO litter dataset for 20,000 iterations, threw in aggressive augmentation (random flips, brightness jitter, ±15° rotations), and ended up with a bounding-box AP50 of 27.60% and a segmentation AP50 of 25.29%, with a mean inference time of 141.59 ms per frame on an NVIDIA T4 GPU. On the front end, a Flutter app lets citizens photograph waste, get disposal advice mapped to BBMP bin categories, and submit GPS-tagged reports. On the back end, a React dashboard gives municipal administrators a dual-mode heatmap with a one-kilometre radius scanner for planning targeted cleanups. The entire backend sits inside a Docker container on Hugging Face Spaces, with MongoDB Atlas handling persistence and Cloudinary hosting images so the database does not choke on base64 blobs.
We set out to build a waste monitoring system that could do four things no existing tool handles together: look at a mixed pile, individually outline each object, tell a citizen which bin to use, and give a municipal planner a spatial map of contamination hotspots. EcoMask does all four.
The Mask R-CNN backbone, fine-tuned on TACO for 20,000 iterations, handles the hard part: pixel-level separation of overlapping objects. The Flutter app handles the human part: making the AI accessible to somebody who has never heard of instance segmentation and just wants to know where to throw their garbage. The Flask backend does the computa-tion, manages authentication, and persists data to MongoDB and Cloudinary. The React dashboard closes the loop by turning individual scan reports into geospatial intelligence: density heatmaps, weekly volume trends, and a one-kilometre radius audit scanner that lets ward officers focus on the worst-affected neighbourhoods.
Are the numbers impressive? Honestly, not by computer vision competition standards. AP50 of 27.60% for boxes and 25.29% for masks is modest. But waste detection is funda-mentally harder than detecting cars or pedestrians. Objects are amorphous. Textures are inconsistent. Occlusion is normal, not exceptional. And the model reliably separates touching objects and produces usable masks under these conditions. For a civic application, that is the result that matters.
References
[1]E. H. Saptaputra, N. Bonafix, and A. S. Araffanda, “Mobile App as Digitalisation of Waste Sorting Management,” in IOP Conf. Ser.: Earth Environ. Sci., vol. 1169, no. 1, p. 012007, 2023. [2]S. Pavithra, S. Aishwarya, and V. A. Senthil, “IoT Based Automated Waste Collection System for Smart Cities,” in Proc. 2023 Int. Conf. Comput. Commun. and Informatics (ICCCI), Coimbatore, India, 2023,
pp. 1–6.
[3]C. Perera et al., “Smart Waste Management System using IoT,” in Proc. 2022 Int. Conf. Adv. Inf. Technol. (ICAIT), Yangon, Myanmar, 2022, pp. 112–118.
[4]V. Ruiz, A´ . Sa´nchez, J. F. Ve´lez, and B. Raducanu, “Automatic Image-Based Waste Classification,” in From Bioinspired Systems and Biomed-ical Applications to Machine Learning, vol. 11487. Cham: Springer, 2019, pp. 422–431.
[5]C. J. Yi and C. F. Kim, “AI-Powered Waste Classification Using Convolutional Neural Networks (CNNs),” Int. J. Adv. Comput. Sci. Appl., vol. 15, no. 10, pp. 67–75, 2024.
[6]A. Dedania, D. Panchal, J. Patel, and J. Jani, “AI-Powered Garbage Classification System with IoT Implementation,” Int. J. Innov. Res. Multidisciplinary Phys. Sci., vol. 13, no. 2, Mar. 2025.
[7]S. G. Fuse, R. N. Kathkoriya, D. A. Lahane, R. S. Nagane, and K. K. Chhajed, “AI-Based Waste Classification & Reporting System,” Int. J. Res. Publ. Rev., vol. 6, no. 4, pp. 13817–13822, Apr. 2025.
[8]S. Majchrowska et al., “Deep learning-based waste detection in natural and urban environments,” Waste Manage., vol. 138, pp. 274–284, Feb. 2022.
[9]A. Alourani, M. S. Al-Shammari, and A. A. Al-Zahrani, “Smart waste management and classification system using advanced IoT and AI technologies,” PeerJ Comput. Sci., vol. 8, p. e1042, 2022.
[10]P. Proenc¸a and P. Simo˜es, “TACO: Trash Annotations in Context for Litter Detection,” GitHub Repository, 2020.
[11]Facebook AI Research (FAIR), “Detectron2 Documentation,” Facebook Research, 2019.
How to Cite This Paper
Aswin T Sunil Kumar, Ms. Geetanjali R (2026). Urban Waste Profiling and Reporting System using Mask R-CNN Model. International Journal of Computer Techniques, 13(5). ISSN: 2394-2231.
